Head-to-head comparison
Claude Opus 5 vs. GPT-5.6 Sol
Compare published benchmark results from matching versions and cohorts, API pricing, and technical specifications. Any documented test setup differences remain visible.
248
shared result series
32
distinct benchmarks
2026-08-16
latest retrieval
Change model selection
Specifications and cost
General specifications and standard API pricing
Undisclosed parameter counts and knowledge cutoffs stay marked as unavailable. The table shows direct standard token pricing, including documented context tiers. Batch, fast, priority, regional surcharges, tool calls, and cloud platform pricing are excluded.
| Attribute | Claude Opus 5 | GPT-5.6 Sol |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Released | Jul 24, 2026 | Jun 26, 2026 |
| Availability | Active | Active |
| Model type | Proprietary | Proprietary |
| Parameters | Not published | Not published |
| Architecture | Not published | Not published |
| Context window | 1,000,000 tokens | 1,050,000 tokens |
| Knowledge cutoff | May 2026 | February 16, 2026 |
| Technical sources | Model, Knowledge cutoff | Model, Context, Knowledge cutoff |
| API pricing per 1M tokens | ||
| API input | $5 | $5 ≤272K / $10 >272K |
| API output | $25 | $30 ≤272K / $45 >272K |
| Cache read | $0.5 | $0.5 ≤272K / $1 >272K |
| Cache write (5 min.) | $6.25 | $6.25 ≤272K / $12.5 >272K |
| Cache write (1 hr.) | $10 | Not listed |
| Price verified | 07/30/2026Source | 08/11/2026Source |
Claude Opus 5GPT-5.6 Sol
API inputUSD per 1M tokens, base tier
Claude Opus 5$5Tie
GPT-5.6 Sol$5Tie
API outputUSD per 1M tokens, base tier
Claude Opus 5$25Lower price
GPT-5.6 Sol$30
Cache readUSD per 1M tokens, base tier
Claude Opus 5$0.5Tie
GPT-5.6 Sol$0.5Tie
Performance
Shared benchmarks
Only published results with the same benchmark version, task, metric, and comparison cohort are paired. Different reasoning levels, agents, harnesses, or output limits appear directly in the table.
248 shared result series
Claude Opus 5GPT-5.6 Sol
AlmanBench v0.1Accepted cases. Higher is better. No documented setup difference
Claude Opus 594.947 %
GPT-5.6 Sol95.141 %Winner
ARC-AGI-3RHAE overall score. Higher is better. No documented setup difference
Claude Opus 530.16 %Winner
GPT-5.6 Sol2.15 %
Arena Text, Style ControlOverall. Higher is better. Test setup not documented as identical
Claude Opus 51,493.745 arena pointsHigher measured value
GPT-5.6 Sol1,480.992 arena points
Code MigrationOverall. Higher is better. No documented setup difference
Claude Opus 557.473 %Winner
GPT-5.6 Sol52.916 %
CorpFin v2Overall. Higher is better. No documented setup difference
Claude Opus 573.194 %Winner
GPT-5.6 Sol64.375 %
CyberBenchOverall. Higher is better. No documented setup difference
Claude Opus 540.678 %
GPT-5.6 Sol88.136 %Winner
EMBOverall. Higher is better. No documented setup difference
Claude Opus 573.556 %Winner
GPT-5.6 Sol72.339 %
Finance Agent (v2)Overall. Higher is better. No documented setup difference
Claude Opus 558.633 %Winner
GPT-5.6 Sol53.756 %
GPQA DiamondOverall. Higher is better. No documented setup difference
Claude Opus 593.43 %
GPT-5.6 Sol95.2 %Winner
Harvey's Legal Agent BenchmarkOverall · Task fully resolved. Higher is better. No documented setup difference
Claude Opus 56.667 %Winner
GPT-5.6 Sol2.5 %
IOIOverall. Higher is better. No documented setup difference
Claude Opus 591.667 %Winner
GPT-5.6 Sol86.667 %
Legal Research BenchOverall · All-pass. Higher is better. No documented setup difference
Claude Opus 555.288 %Winner
GPT-5.6 Sol48.077 %
| Benchmark and task | Claude Opus 5 | GPT-5.6 Sol | Comparability | Source |
|---|---|---|---|---|
| AlmanBench v0.1Accepted casesReasoning: max | 94.947 % | 95.141 %Winner | v0.1, results commit f34ddbfaccepted casesHigher is betterCohort: almanbench-v0-1:official:overallNo documented setup differenceWinner: GPT-5.6 SolClaude Opus 5: Reasoning: max GPT-5.6 Sol: Reasoning: max1,029 public tasks with one direct run per row. The target is the Alman language specification. | Alman InstitutType: Independent evaluationLeaderboard updated: 2026-08-05Import ID: curated-alman-institut-2026-08-09-dae32c26388f / curated-alman-institut-2026-08-09-c025c420dbf4Retrieved 2026-08-09 |
| ARC-AGI-3RHAE overall scoreReasoning: high | 30.16 %Winner | 2.15 % | ARC-AGI-3 2026 verifiedRHAE scoreHigher is betterCohort: arc-agi-3-semi-private-verified-official-harnessNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: Reasoning: high, Agent: ARC Prize official harness, Harness: ARC Prize official harness GPT-5.6 Sol: Reasoning: high, Agent: ARC Prize official harness, Harness: ARC Prize official harnessOfficial ARC Prize harness on unseen interactive environments. Reasoning levels remain separate. | ARC Prize, ARC PrizeType: Official benchmark leaderboardImport ID: curated-arc-prize-2026-08-06-efa794b86f9e / curated-arc-prize-2026-08-06-7f3603da855aRetrieved 2026-08-06 |
| Arena Text, Style ControlOverall | 1,493.745 arena pointsHigher measured value 95% confidence interval: ±5.332The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result. | 1,480.992 arena points 95% confidence interval: ±5.733The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result. | style-control-v1style-controlled Bradley-Terry ratingHigher is betterCohort: lmarena-text-style-control:2026-08-12:overallTest setup not documented as identicalNo direct winnerClaude Opus 5: Snapshot: claude-opus-5-high, Reasoning: high, Harness: Arena text, style control GPT-5.6 Sol: Snapshot: gpt-5.6-sol-xhigh, Reasoning: xhigh, Harness: Arena text, style controlClaude Opus 5: Style-controlled Arena rating from 19,792 anonymous pairwise votes.GPT-5.6 Sol: Style-controlled Arena rating from 15,450 anonymous pairwise votes. | ArenaType: Official benchmark leaderboardLeaderboard updated: 2026-08-12Import ID: lmarena-eeedf9209373a7f9b324bfb839e7b02fedacd1e5-1611ed64f3f1Retrieved 2026-08-16 |
| Code MigrationOverall | 57.473 %Winner | 52.916 % | 1accuracyHigher is betterCohort: vals-code-migration:1:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-13Import ID: vals-code-migration-2026-08-16-107652dc6bfcRetrieved 2026-08-16 |
| CorpFin v2Overall | 73.194 %Winner | 64.375 % | 2accuracyHigher is betterCohort: vals-corp-fin-v2:2:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-12Import ID: vals-corp_fin_v2-2026-08-16-45bbf84f57abRetrieved 2026-08-16 |
| CyberBenchOverall | 40.678 % | 88.136 %Winner | 1accuracyHigher is betterCohort: vals-cyberbench:1:overallNo documented setup differenceWinner: GPT-5.6 SolClaude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-03Import ID: vals-cyber-2026-08-16-702acaee9aa9Retrieved 2026-08-16 |
| EMBOverall | 73.556 %Winner | 72.339 % | 1accuracyHigher is betterCohort: vals-emb:1:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-13Import ID: vals-emb-2026-08-16-68d5d1d47a97Retrieved 2026-08-16 |
| Finance Agent (v2)Overall | 58.633 %Winner | 53.756 % | 2accuracyHigher is betterCohort: vals-finance-agent-v2:2:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-14Import ID: vals-fabv2-2026-08-16-8853da43085dRetrieved 2026-08-16 |
| GPQA DiamondOverall | 93.43 % Small gaps are not automatically meaningful because the benchmark is saturated and noisy. | 95.2 %Winner Small gaps are not automatically meaningful because the benchmark is saturated and noisy. | DiamondaccuracyHigher is betterCohort: vals-gpqa-diamond-2026-08-01No documented setup differenceWinner: GPT-5.6 SolClaude Opus 5: Attempts: pass@1 GPT-5.6 Sol: Attempts: pass@1Shared Vals evaluation of the largely saturated benchmark. | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-01Import ID: curated-vals-ai-2026-08-02-71e39d20f05d / curated-vals-ai-2026-08-02-40bfd78b0ea4Retrieved 2026-08-02 |
| Harvey's Legal Agent BenchmarkOverall · Task fully resolved | 6.667 %Winner | 2.5 % | 1task resolution rateHigher is betterCohort: vals-legal-agent-benchmark:1:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-14Import ID: vals-hlab-2026-08-16-bae90566df2fRetrieved 2026-08-16 |
| IOIOverall | 91.667 %Winner | 86.667 % | 1accuracyHigher is betterCohort: vals-ioi:1:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-09Import ID: vals-ioi-2026-08-16-d9164ef483aaRetrieved 2026-08-16 |
| Legal Research BenchOverall · All-pass | 55.288 %Winner | 48.077 % | 1all-pass rateHigher is betterCohort: vals-legal-research:1:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-14Import ID: vals-legal_research-2026-08-16-af296c00ecaaRetrieved 2026-08-16 |
| LegalBenchOverall | 86.974 %Winner | 86.965 % | 1accuracyHigher is betterCohort: vals-legal-bench:1:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-12Import ID: vals-legal_bench-2026-08-16-d5b3b1b2b5c8Retrieved 2026-08-16 |
| LiveCodeBench v6Overall | 89.033 %Winner | 82.604 % | 1accuracyHigher is betterCohort: vals-lcb:1:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-13Import ID: vals-lcb-2026-08-16-17b9c147080dRetrieved 2026-08-16 |
| MedCodeOverall | 63.57 %Winner | 43.974 % | 1accuracyHigher is betterCohort: vals-medcode:1:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-12Import ID: vals-medcode-2026-08-16-524f7f474eacRetrieved 2026-08-16 |
| MedScribeOverall | 90.985 %Winner | 85.233 % | 1accuracyHigher is betterCohort: vals-medscribe:1:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-12Import ID: vals-medscribe-2026-08-16-c385354a7e62Retrieved 2026-08-16 |
| MMLU-ProOverall | 91.591 %Winner | 89.1 % | 1accuracyHigher is betterCohort: vals-mmlu-pro:1:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-12Import ID: vals-mmlu_pro-2026-08-16-e14d5bb3a34cRetrieved 2026-08-16 |
| MMMU-ProOverall | 89.88 %Winner | 88.84 % | ProaccuracyHigher is betterCohort: vals-mmmu-pro-pass1-temp0-2026-07-31No documented setup differenceWinner: Claude Opus 5Claude Opus 5: Attempts: pass@1 GPT-5.6 Sol: Attempts: pass@1Approximately 1,700 multimodal tasks, Pass@1 at temperature 0. | Vals AIType: Independent evaluationLeaderboard updated: 2026-07-31Import ID: curated-vals-ai-2026-08-02-7cb2c9e11cfa / curated-vals-ai-2026-08-02-077b07f92fa3Retrieved 2026-08-02 |
| MortgageTaxOverall | 72.059 %Winner | 67.289 % | 1accuracyHigher is betterCohort: vals-mortgage-tax:1:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-12Import ID: vals-mortgage_tax-2026-08-16-e78c985dbb1cRetrieved 2026-08-16 |
| ProgramBenchOverall | 3 %Winner | 1.5 % | 1accuracyHigher is betterCohort: vals-programbench:1:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-13Import ID: vals-programbench-2026-08-16-08d9727662ddRetrieved 2026-08-16 |
| ProofBenchOverall | 78 %Winner | 77 % | currentsolved proofsHigher is betterCohort: vals-proofbench-lean4-40-steps-2026-07-31No documented setup differenceWinner: Claude Opus 5Claude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed200 Lean 4 problems with tool access and up to 40 steps. | Vals AIType: Independent evaluationLeaderboard updated: 2026-07-31Import ID: curated-vals-ai-2026-08-02-0c2bc33f9e1f / curated-vals-ai-2026-08-02-35b7a2f5cfe7Retrieved 2026-08-02 |
| Public Benefits Bench v1.1Overall | 76.928 %Winner | 66.509 % | 1.1accuracyHigher is betterCohort: vals-snap:1.1:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-12Import ID: vals-public-benefits-bench-2026-08-16-7ce68123c502Retrieved 2026-08-16 |
| ReverseEngBenchOverall | 12.214 % | 30.534 %Winner | 1accuracyHigher is betterCohort: vals-sre-bench:1:overallNo documented setup differenceWinner: GPT-5.6 SolClaude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-10Import ID: vals-reverse_eng-2026-08-16-625e75d80df9Retrieved 2026-08-16 |
| SAGEOverall | 49.426 % | 52.563 %Winner | 1accuracyHigher is betterCohort: vals-sage:1:overallNo documented setup differenceWinner: GPT-5.6 SolClaude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-12Import ID: vals-sage-2026-08-16-a416dd25affbRetrieved 2026-08-16 |
| SkillsBenchOverall | 60.441 %Winner | 54.1 % | 1accuracyHigher is betterCohort: vals-skillsbench:1:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: Harness: OpenHands GPT-5.6 Sol: Harness: OpenHands | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-13Import ID: vals-skillsbench-2026-08-16-4917a67268d2Retrieved 2026-08-16 |
| SWE-bench VerifiedOverall | 97 %Winner Editorial selection from the Vals AI table, not a complete extract of the comparison cohort. | 96.2 % Editorial selection from the Vals AI table, not a complete extract of the comparison cohort. | VerifiedresolvedHigher is betterCohort: swe-verified-vals-mini-swe-agent-2026-07-31No documented setup differenceWinner: Claude Opus 5Claude Opus 5: Agent: mini-SWE-agent, Harness: mini-SWE-agent GPT-5.6 Sol: Agent: mini-SWE-agent, Harness: mini-SWE-agentShared mini-SWE-agent harness with bash tools. | Vals AIType: Independent evaluationLeaderboard updated: 2026-07-31Import ID: curated-vals-ai-2026-08-01-6d188ba56b5b / curated-vals-ai-2026-08-01-078dcd2c807cRetrieved 2026-08-01 |
| TaxEval v2Overall | 75.143 %Winner | 74.775 % | 2accuracyHigher is betterCohort: vals-tax-eval-v2:2:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-12Import ID: vals-tax_eval_v2-2026-08-16-5bbd495dac53Retrieved 2026-08-16 |
| Terminal-Bench 2.1Overall | 84.64 % Nine passing tasks used an Opus 4.8 fallback because of safety systems. Counting them as failures lowers the Vals result to 81.27%. | 85.77 %Winner | 2.1pass@1Higher is betterCohort: vals-terminal-bench-2-1-terminus2-2026-08-01No documented setup differenceWinner: GPT-5.6 SolClaude Opus 5: Agent: Terminus 2, Harness: Terminus 2, Attempts: pass@1 GPT-5.6 Sol: Agent: Terminus 2, Harness: Terminus 2, Attempts: pass@1Constant Terminus 2 agent, 89 tasks, Daytona environments, and a shared time limit. | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-01Import ID: curated-vals-ai-2026-08-02-52c225d36f55 / curated-vals-ai-2026-08-02-db4549f6f7c5Retrieved 2026-08-02 |
| Time Horizon Index: KSPOverall | 13.667 %Winner | 13 % | 1progressHigher is betterCohort: vals-ksp-bench:1:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-05Import ID: vals-time_horizon_index-2026-08-16-b63dc486165aRetrieved 2026-08-16 |
| Vals IndexOverall | 67.213 %Winner | 63.709 % | 2weighted index scoreHigher is betterCohort: vals-vals-index:2:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-14Import ID: vals-vals_index-2026-08-16-ab641d89d6edRetrieved 2026-08-16 |
| Vals Multimodal IndexOverall | 73.896 %Winner | 72.64 % | 1.2weighted index scoreHigher is betterCohort: vals-vals-multimodal-index:1.2:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-11Import ID: vals-vals_multimodal_index-2026-08-16-9f3ebb221408Retrieved 2026-08-16 |
| Vibe Code Bench v1.1Overall | 88.404 %Winner | 80.495 % | 1.1accuracyHigher is betterCohort: vals-vibe-code-bench:1.1:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: Harness: OpenHands GPT-5.6 Sol: Harness: OpenHands | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-13Import ID: vals-vibe-code-2026-08-16-a73c9770bcc4Retrieved 2026-08-16 |
| ARC-AGI-3RHAE overall score | 30.16 %Higher measured value | 0.33 % | ARC-AGI-3 2026 verifiedRHAE scoreHigher is betterCohort: arc-agi-3-semi-private-verified-official-harnessTest setup not documented as identicalNo direct winnerClaude Opus 5: Reasoning: high, Agent: ARC Prize official harness, Harness: ARC Prize official harness GPT-5.6 Sol: Reasoning: low, Agent: ARC Prize official harness, Harness: ARC Prize official harnessOfficial ARC Prize harness on unseen interactive environments. Reasoning levels remain separate. | ARC Prize, ARC PrizeType: Official benchmark leaderboardImport ID: curated-arc-prize-2026-08-06-efa794b86f9e / curated-arc-prize-2026-08-06-691198a6be3bRetrieved 2026-08-06 |
| Arena Text, Style ControlOverall | 1,491.094 arena pointsHigher measured value 95% confidence interval: ±6.993The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result. | 1,480.992 arena points 95% confidence interval: ±5.733The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result. | style-control-v1style-controlled Bradley-Terry ratingHigher is betterCohort: lmarena-text-style-control:2026-08-12:overallTest setup not documented as identicalNo direct winnerClaude Opus 5: Snapshot: claude-opus-5-max, Reasoning: max, Harness: Arena text, style control GPT-5.6 Sol: Snapshot: gpt-5.6-sol-xhigh, Reasoning: xhigh, Harness: Arena text, style controlClaude Opus 5: Style-controlled Arena rating from 9,559 anonymous pairwise votes.GPT-5.6 Sol: Style-controlled Arena rating from 15,450 anonymous pairwise votes. | ArenaType: Official benchmark leaderboardLeaderboard updated: 2026-08-12Import ID: lmarena-eeedf9209373a7f9b324bfb839e7b02fedacd1e5-1611ed64f3f1Retrieved 2026-08-16 |
| Code MigrationCLI | 53.297 %Higher measured value | 47.221 % | 1accuracyHigher is betterCohort: vals-code-migration:1:cliTest setup not documented as identicalNo direct winnerClaude Opus 5: No further settings disclosed GPT-5.6 Sol: Reasoning: max, Output limit: 128,000 tokens | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-09Import ID: vals-code-migration-2026-08-11-3a2d5ce718a3Retrieved 2026-08-11 |
| CorpFin v2Exact Pages | 68.648 %Higher measured value | 61.538 % | 2accuracyHigher is betterCohort: vals-corp-fin-v2:2:exact_pagesTest setup not documented as identicalNo direct winnerClaude Opus 5: No further settings disclosed GPT-5.6 Sol: Reasoning: max, Output limit: 128,000 tokens | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-06Import ID: vals-corp_fin_v2-2026-08-11-d9e58d191cd9Retrieved 2026-08-11 |
| CyberBenchPatch | 81.356 % | 83.051 %Winner | 1accuracyHigher is betterCohort: vals-cyberbench:1:patchNo documented setup differenceWinner: GPT-5.6 SolClaude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-03Import ID: vals-cyber-2026-08-11-1f20498a3dc1Retrieved 2026-08-11 |
| EMBComps | 75.171 %Higher measured value | 71.309 % | 1accuracyHigher is betterCohort: vals-emb:1:compsTest setup not documented as identicalNo direct winnerClaude Opus 5: No further settings disclosed GPT-5.6 Sol: Reasoning: max, Output limit: 128,000 tokens | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-06Import ID: vals-emb-2026-08-11-e3c855e236e4Retrieved 2026-08-11 |
| Finance Agent (v2)Adjustments | 54.617 %Higher measured value | 49.233 % | 2accuracyHigher is betterCohort: vals-finance-agent-v2:2:adjustmentsTest setup not documented as identicalNo direct winnerClaude Opus 5: Compute: max, Temperature: 1, Output limit: 128,000 tokens GPT-5.6 Sol: Reasoning: max, Output limit: 128,000 tokens | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-06Import ID: vals-fabv2-2026-08-11-3e5c93a1857cRetrieved 2026-08-11 |
| GPQA DiamondOverall | 93.434 % | 95.202 %Winner | 1accuracyHigher is betterCohort: vals-gpqa:1:overallNo documented setup differenceWinner: GPT-5.6 SolClaude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-12Import ID: vals-gpqa-2026-08-16-7f8363eaa2dbRetrieved 2026-08-16 |
| Harvey's Legal Agent BenchmarkAntitrust Competition · Criteria passed | 87.319 %Higher measured value | 78.623 % | 1criteria pass rateHigher is betterCohort: vals-legal-agent-benchmark:1:criteria_pass_rate__antitrust_competitionTest setup not documented as identicalNo direct winnerClaude Opus 5: No further settings disclosed GPT-5.6 Sol: Reasoning: max, Output limit: 128,000 tokens | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-07Import ID: vals-hlab-2026-08-11-ea11b6e53414Retrieved 2026-08-11 |
| IOIIOI 2024 | 100 %Higher measured value | 85 % | 1accuracyHigher is betterCohort: vals-ioi:1:ioi2024Test setup not documented as identicalNo direct winnerClaude Opus 5: Compute: max, Temperature: 1, Output limit: 128,000 tokens GPT-5.6 Sol: Reasoning: max, Output limit: 128,000 tokens | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-09Import ID: vals-ioi-2026-08-11-2dfe8b4b8b7bRetrieved 2026-08-11 |
| Legal Research BenchAdministrative / Regulatory · All-pass | 65.152 %Higher measured value | 60.606 % | 1all-pass rateHigher is betterCohort: vals-legal-research:1:administrative_regulatoryTest setup not documented as identicalNo direct winnerClaude Opus 5: No further settings disclosed GPT-5.6 Sol: Reasoning: max, Output limit: 128,000 tokens | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-06Import ID: vals-legal_research-2026-08-11-326cf7aaca58Retrieved 2026-08-11 |
| LegalBenchConclusion Tasks | 91.56 %Higher measured value | 90.492 % | 1accuracyHigher is betterCohort: vals-legal-bench:1:conclusion_tasksTest setup not documented as identicalNo direct winnerClaude Opus 5: No further settings disclosed GPT-5.6 Sol: Reasoning: max, Output limit: 128,000 tokens | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-09Import ID: vals-legal_bench-2026-08-11-836e5e44618eRetrieved 2026-08-11 |
| LiveCodeBench v6Easy | 99.068 %Higher measured value | 98.137 % | 1accuracyHigher is betterCohort: vals-lcb:1:easyTest setup not documented as identicalNo direct winnerClaude Opus 5: No further settings disclosed GPT-5.6 Sol: Reasoning: max, Output limit: 128,000 tokens | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-09Import ID: vals-lcb-2026-08-11-0342bcc364e4Retrieved 2026-08-11 |
| MMLU-ProBiology | 95.676 %Higher measured value | 94.561 % | 1accuracyHigher is betterCohort: vals-mmlu-pro:1:biologyTest setup not documented as identicalNo direct winnerClaude Opus 5: No further settings disclosed GPT-5.6 Sol: Reasoning: max, Output limit: 128,000 tokens | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-09Import ID: vals-mmlu_pro-2026-08-11-b8e53828d746Retrieved 2026-08-11 |
| MMMU-ProOverall | 89.884 %Winner | 88.844 % | 1accuracyHigher is betterCohort: vals-mmmu:1:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-09Import ID: vals-mmmu-2026-08-16-37c9a8c7e21fRetrieved 2026-08-16 |
| MortgageTaxNumerical Extraction | 81.399 %Higher measured value | 75.119 % | 1accuracyHigher is betterCohort: vals-mortgage-tax:1:numerical_extractionTest setup not documented as identicalNo direct winnerClaude Opus 5: No further settings disclosed GPT-5.6 Sol: Reasoning: max, Output limit: 128,000 tokens | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-09Import ID: vals-mortgage_tax-2026-08-11-5ea6eef3e482Retrieved 2026-08-11 |
| ProgramBenchAlmost Resolved | 41.5 %Higher measured value | 23 % | 1accuracyHigher is betterCohort: vals-programbench:1:almostTest setup not documented as identicalNo direct winnerClaude Opus 5: No further settings disclosed GPT-5.6 Sol: Reasoning: max, Output limit: 128,000 tokens | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-09Import ID: vals-programbench-2026-08-11-b6f8959fbdfeRetrieved 2026-08-11 |
| ProofBenchOverall | 99 %Winner | 83 % | 1.1accuracyHigher is betterCohort: vals-proof-bench:1.1:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-14Import ID: vals-proof_bench-2026-08-16-14232db24de1Retrieved 2026-08-16 |
| SAGEAutomata | 32.54 % | 39.682 %Higher measured value | 1accuracyHigher is betterCohort: vals-sage:1:automataTest setup not documented as identicalNo direct winnerClaude Opus 5: Temperature: 1, Output limit: 30,000 tokens GPT-5.6 Sol: Reasoning: max, Output limit: 30,000 tokens | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-09Import ID: vals-sage-2026-08-11-f1f79f6d8af3Retrieved 2026-08-11 |
| SkillsBenchNo Skills | 45.191 %Higher measured value | 43.312 % | 1accuracyHigher is betterCohort: vals-skillsbench:1:no_skillsTest setup not documented as identicalNo direct winnerClaude Opus 5: Compute: max, Harness: OpenHands, Temperature: 1, Output limit: 128,000 tokens GPT-5.6 Sol: Reasoning: max, Harness: OpenHands, Output limit: 128,000 tokens | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-07Import ID: vals-skillsbench-2026-08-11-351bec9d0e01Retrieved 2026-08-11 |
| SWE-bench VerifiedOverall | 97 %Winner | 96.2 % | 1accuracyHigher is betterCohort: vals-swebench:1:overallNo documented setup differenceWinner: Claude Opus 5Claude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-12Import ID: vals-swebench-2026-08-16-c133071b8167Retrieved 2026-08-16 |
| TaxEval v2Correctness | 61.979 % | 65.004 %Higher measured value | 2accuracyHigher is betterCohort: vals-tax-eval-v2:2:correctnessTest setup not documented as identicalNo direct winnerClaude Opus 5: No further settings disclosed GPT-5.6 Sol: Reasoning: max, Output limit: 128,000 tokens | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-09Import ID: vals-tax_eval_v2-2026-08-11-d05a40bf446fRetrieved 2026-08-11 |
| Terminal-Bench 2.1Overall | 84.644 % | 85.768 %Winner | 2.1accuracyHigher is betterCohort: vals-terminal-bench-2-1:2.1:overallNo documented setup differenceWinner: GPT-5.6 SolClaude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-13Import ID: vals-terminal-bench-2-1-2026-08-16-b3f38c18bea4Retrieved 2026-08-16 |
| Time Horizon Index: KSPProgress efficiency | 11.947 %Higher measured value | 10.634 % | 1progressHigher is betterCohort: vals-ksp-bench:1:mean_progress_to_dateTest setup not documented as identicalNo direct winnerClaude Opus 5: Compute: max, Temperature: 1, Output limit: 128,000 tokens GPT-5.6 Sol: Reasoning: max, Output limit: 128,000 tokens | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-05Import ID: vals-time_horizon_index-2026-08-11-f0a20a098b28Retrieved 2026-08-11 |
| Vals IndexCorpFin v2 | 74.942 %Higher measured value | 66.434 % | 1.2benchmark scoreHigher is betterCohort: vals-vals-index:1.2:corp_fin_v2Test setup not documented as identicalNo direct winnerClaude Opus 5: No further settings disclosed GPT-5.6 Sol: Reasoning: max, Output limit: 128,000 tokens | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-09Import ID: vals-vals_index-2026-08-11-e79d08d59249Retrieved 2026-08-11 |
| Vals Multimodal IndexCorpFin v2 | 74.942 %Higher measured value | 66.434 % | 1.2benchmark scoreHigher is betterCohort: vals-vals-multimodal-index:1.2:corp_fin_v2Test setup not documented as identicalNo direct winnerClaude Opus 5: No further settings disclosed GPT-5.6 Sol: Reasoning: max, Output limit: 128,000 tokens | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-09Import ID: vals-vals_multimodal_index-2026-08-11-fa704862df19Retrieved 2026-08-11 |
| ARC-AGI-3RHAE overall score | 30.16 %Higher measured value | 7.78 % | ARC-AGI-3 2026 verifiedRHAE scoreHigher is betterCohort: arc-agi-3-semi-private-verified-official-harnessTest setup not documented as identicalNo direct winnerClaude Opus 5: Reasoning: high, Agent: ARC Prize official harness, Harness: ARC Prize official harness GPT-5.6 Sol: Reasoning: max, Agent: ARC Prize official harness, Harness: ARC Prize official harnessOfficial ARC Prize harness on unseen interactive environments. Reasoning levels remain separate. | ARC Prize, ARC PrizeType: Official benchmark leaderboardImport ID: curated-arc-prize-2026-08-06-efa794b86f9e / curated-arc-prize-2026-08-06-9ebb2edeedb6Retrieved 2026-08-06 |
| Arena Text, Style ControlBusiness, management and finance | 1,488.821 arena pointsHigher measured value 95% confidence interval: ±10.372The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result. | 1,482.723 arena points 95% confidence interval: ±11.356The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result. | style-control-v1style-controlled Bradley-Terry ratingHigher is betterCohort: lmarena-text-style-control:2026-08-12:industry_business_and_management_and_financial_operationsTest setup not documented as identicalNo direct winnerClaude Opus 5: Snapshot: claude-opus-5-high, Reasoning: high, Harness: Arena text, style control GPT-5.6 Sol: Snapshot: gpt-5.6-sol-xhigh, Reasoning: xhigh, Harness: Arena text, style controlClaude Opus 5: Style-controlled Arena rating from 3,811 anonymous pairwise votes.GPT-5.6 Sol: Style-controlled Arena rating from 3,027 anonymous pairwise votes. | ArenaType: Official benchmark leaderboardLeaderboard updated: 2026-08-12Import ID: lmarena-eeedf9209373a7f9b324bfb839e7b02fedacd1e5-1611ed64f3f1Retrieved 2026-08-16 |
| Code MigrationCOBOL | 70 %No direct winner | 70 %No direct winner | 1accuracyHigher is betterCohort: vals-code-migration:1:cobolTest setup not documented as identicalNo direct winnerClaude Opus 5: No further settings disclosed GPT-5.6 Sol: Reasoning: max, Output limit: 128,000 tokens | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-09Import ID: vals-code-migration-2026-08-11-3a2d5ce718a3Retrieved 2026-08-11 |
| CorpFin v2Max Fitting Context | 75.991 %Higher measured value | 65.152 % | 2accuracyHigher is betterCohort: vals-corp-fin-v2:2:max_fitting_contextTest setup not documented as identicalNo direct winnerClaude Opus 5: No further settings disclosed GPT-5.6 Sol: Reasoning: max, Output limit: 128,000 tokens | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-06Import ID: vals-corp_fin_v2-2026-08-11-d9e58d191cd9Retrieved 2026-08-11 |
| CyberBenchPoC | 0 % | 93.22 %Winner | 1accuracyHigher is betterCohort: vals-cyberbench:1:pocNo documented setup differenceWinner: GPT-5.6 SolClaude Opus 5: No further settings disclosed GPT-5.6 Sol: No further settings disclosed | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-03Import ID: vals-cyber-2026-08-11-1f20498a3dc1Retrieved 2026-08-11 |
| EMBDataroom | 87.681 %Higher measured value | 85.243 % | 1accuracyHigher is betterCohort: vals-emb:1:dataroomTest setup not documented as identicalNo direct winnerClaude Opus 5: No further settings disclosed GPT-5.6 Sol: Reasoning: max, Output limit: 128,000 tokens | Vals AIType: Independent evaluationLeaderboard updated: 2026-08-06Import ID: vals-emb-2026-08-11-e3c855e236e4Retrieved 2026-08-11 |
The table shows 64 diversified result series from 248 shared results. This prevents one benchmark with many subtasks from crowding out every other test.
How to read this comparison
Winning one benchmark is not an overall verdict
Your workload, budget, and required context length matter most. A coding benchmark says little about visual understanding or agent performance.
Prices use official API rates per 1M tokens. Web subscriptions and cloud platform surcharges can differ.
Every result links to its measurement source and retrieval date. Labels distinguish vendor reports, official benchmark leaderboards, and independent evaluations. Disputed or archived results are excluded.