GPT-5.3-Codex
Unavailable in this round
Language modelDeprecated
GPT-5.3-Codex
OpenAI
- Released
- February 5, 2026
- Data date
- September 8, 2026
- Predecessor
- GPT-5.2-Codex (Official documentation)
- Speed
- 139.4 tokens/sOutput throughput
- Price
- Input · Output / 1M tokens
- Input
- Text, Image
- Output
- Text
OpenAI describes GPT-5.3-Codex as its first model to play a substantial role in its own development. Early versions helped the team debug training, manage deployment, and analyze tests. It combines GPT-5.2-Codex’s coding capabilities with GPT-5.2’s reasoning and professional knowledge.
Those coding capabilities later fed into GPT-5.4, which also supports native computer use. OpenAI deprecated GPT-5.3-Codex on October 1, 2026, and announced shutdown for April 1, 2027.
Specifications and access
| Specification | Value and source |
|---|---|
| Weights | ProprietarySource |
| Access | APISource |
| Context window | 400,000 tokensSource |
| Knowledge cutoff | August 31, 2025Source |
| Notes | Combines GPT-5.2-Codex coding ability with GPT-5.2 reasoning, 25% fasterSource |
| API model ID | gpt-5.3-codexSource |
| Maximum output | 128,000 tokensSource |
| Input | Text, ImageSource |
| Output | TextSource |
| Reasoning mode | low, medium, high, xhighSource |
| API interfaces | Responses API, Chat Completions APISource |
| Tool calling | YesSource |
| Structured output | YesSource |
| Release | 2026-02-05Source |
Page 1 of 3
Published benchmarks
Agent tasks
METR Time Horizon
Human task duration at which a model-agent system reaches a 50% or 80% predicted success probability. Minutes refer to human task duration, not model runtime. METR considers estimates above 16 hours unreliable.
| Sub-task | GPT-5.3-Codex (This model) | Gemini 3.1 Pro Preview | GPT-5 | o3 |
|---|---|---|---|---|
| 50% time horizon | 349.53 | 384.15 (best value in this row) | 203.01 | 119.73 |
| 80% time horizon | 54.74 | 89.8 (best value in this row) | 38.31 | 29.98 |
Unit: minuteshigher is betterAs of August 26, 2026Source: METRbars scaled within each row, not across rows
Coding
LiveCodeBench v6
| Sub-task | GPT-5.3-Codex (This model) | GPT-5.4 | GPT-5.5 | GPT-5.4 mini | GPT-5.6 Terra |
|---|---|---|---|---|---|
| Easy | 99.07% (best value in this row) | 99.07% (best value in this row) | 98.45% | 97.83% | 97.21% |
| Hard | 75.14% (best value in this row) | 67.71% | 69.71% | 61.71% | 72.86% |
| Medium | 87.73% (best value in this row) | 85.64% | 87.73% (best value in this row) | 84.86% | 87.73% (best value in this row) |
| Overall | 87.31% (best value in this row) | 84.14% | 85.3% | 81.47% | 85.93% |
higher is betterAs of October 15, 2026Source: Vals AIbars scaled within each row, not across rows
Methodology and context
- Metric
- accuracy
SWE-bench Verified
| Sub-task | GPT-5.3-Codex (This model) | GPT-5.5 | GPT-5.4 | GPT-5.4 mini |
|---|---|---|---|---|
| <15 min fix | 90.21% | 91.75% (best value in this row) | 88.14% | 85.57% |
| >4 hours | 33.33% | 66.67% (best value in this row) | 0% | 0% |
| 1-4 hours | 54.76% (best value in this row) | 50% | 50% | 33.33% |
| 15 min - 1 hour | 73.18% | 81.23% (best value in this row) | 76.25% | 70.88% |
| Overall | 78% | 82.6% (best value in this row) | 78.2% | 73% |
higher is betterAs of October 15, 2026Source: Vals AIbars scaled within each row, not across rows
Methodology and context
- Metric
- accuracy
Recorded measurement details
METR Time Horizon: 50% time horizon (Show measurement, test conditions, and source)
- Source value
- 349.530732
- Score
- 349.530732
- Metric
- 50% time horizon
- Unit
- minutes
- Benchmark version
- TH 1.1
- Category
- agentic
- Task
- p50
- Task label
- 50% time horizon
- Direction
- Higher is better
- Comparison cohort
- metr-th1-1-p50
- Source type
- independent-evaluator
- Evaluator
- METR
- Status
- active
- Retrieved at
- 2026-08-26
- Methodology
- Triframe agent with six independent runs per task. Time is estimated human completion duration.. Source model ID: curated/gpt-5.3-codex. Source model name: GPT-5.3-Codex. Provider reported by source: OpenAI. Model snapshot: gpt-5.3-codex-2026-02-05. Evaluator: METR. Dataset type: public. Leaderboard date: 2026-05-08. Harness: Triframe. Agent: Triframe. Attempts: 6 independent runs per task. Score display: 5.8 h (95% confidence interval 3.2 h to 13.6 h). Source ID: metr. Evidence capture ID: curated-metr-2026-08-26-eae972724332. Source payload hash: eae972724332b484893349fcc227a9b865d78efcd94e3b78d2d11cbc2315801e. Parser version: curated-v1. Imported at: 2026-10-15
- Comparison methodology
- Triframe agent with six independent runs per task. Time is estimated human completion duration.
METR Time Horizon: 80% time horizon (Show measurement, test conditions, and source)
- Source value
- 54.739407
- Score
- 54.739407
- Metric
- 80% time horizon
- Unit
- minutes
- Benchmark version
- TH 1.1
- Category
- agentic
- Task
- p80
- Task label
- 80% time horizon
- Direction
- Higher is better
- Comparison cohort
- metr-th1-1-p80
- Source type
- independent-evaluator
- Evaluator
- METR
- Status
- active
- Retrieved at
- 2026-08-26
- Methodology
- Triframe agent with six independent runs per task. Time is estimated human completion duration.. Source model ID: curated/gpt-5.3-codex. Source model name: GPT-5.3-Codex. Provider reported by source: OpenAI. Model snapshot: gpt-5.3-codex-2026-02-05. Evaluator: METR. Dataset type: public. Leaderboard date: 2026-05-08. Harness: Triframe. Agent: Triframe. Attempts: 6 independent runs per task. Score display: 54.7 min (95% confidence interval 22.4 min to 2 h). Source ID: metr. Evidence capture ID: curated-metr-2026-08-26-bdb68f137fa5. Source payload hash: bdb68f137fa51fbbd7af0345aacd4463b88a571dc43361afdd115bbe86d87dc2. Parser version: curated-v1. Imported at: 2026-10-15
- Comparison methodology
- Triframe agent with six independent runs per task. Time is estimated human completion duration.
LiveCodeBench v6: Easy (Show measurement, test conditions, and source)
- Source value
- 99.068
- Score
- 99.068
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 1
- Category
- coding
- Task
- easy
- Task label
- Easy
- Direction
- Higher is better
- Comparison cohort
- vals-lcb:1:easy
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Source model ID: openai/gpt-5.3-codex. Source model name: GPT-5.3-Codex. Provider reported by source: OpenAI. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-09-01. Reasoning effort: xhigh. Maximum output tokens: 128000. Latency in seconds: 17.469. Standard error: 0.535. Cost per task in USD: 0.013703. Source ID: vals-ai. Evidence capture ID: vals-lcb-2026-10-15-1a7c7dabc9d9. Source payload hash: 1a7c7dabc9d95a0182886b1edc42623987cac2095a0d407a5e334f37acdfa2f5. Parser version: vals-astro-v3. Imported at: 2026-10-15
- Comparison methodology
- Reasoning effort: xhigh. Maximum output tokens: 128000
LiveCodeBench v6: Hard (Show measurement, test conditions, and source)
- Source value
- 75.143
- Score
- 75.143
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 1
- Category
- coding
- Task
- hard
- Task label
- Hard
- Direction
- Higher is better
- Comparison cohort
- vals-lcb:1:hard
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Source model ID: openai/gpt-5.3-codex. Source model name: GPT-5.3-Codex. Provider reported by source: OpenAI. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-09-01. Reasoning effort: xhigh. Maximum output tokens: 128000. Latency in seconds: 306.379. Standard error: 2.31. Cost per task in USD: 0.209533. Source ID: vals-ai. Evidence capture ID: vals-lcb-2026-10-15-1a7c7dabc9d9. Source payload hash: 1a7c7dabc9d95a0182886b1edc42623987cac2095a0d407a5e334f37acdfa2f5. Parser version: vals-astro-v3. Imported at: 2026-10-15
- Comparison methodology
- Reasoning effort: xhigh. Maximum output tokens: 128000
LiveCodeBench v6: Medium (Show measurement, test conditions, and source)
- Source value
- 87.728
- Score
- 87.728
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 1
- Category
- coding
- Task
- medium
- Task label
- Medium
- Direction
- Higher is better
- Comparison cohort
- vals-lcb:1:medium
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Source model ID: openai/gpt-5.3-codex. Source model name: GPT-5.3-Codex. Provider reported by source: OpenAI. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-09-01. Reasoning effort: xhigh. Maximum output tokens: 128000. Latency in seconds: 74.045. Standard error: 1.677. Cost per task in USD: 0.051518. Source ID: vals-ai. Evidence capture ID: vals-lcb-2026-10-15-1a7c7dabc9d9. Source payload hash: 1a7c7dabc9d95a0182886b1edc42623987cac2095a0d407a5e334f37acdfa2f5. Parser version: vals-astro-v3. Imported at: 2026-10-15
- Comparison methodology
- Reasoning effort: xhigh. Maximum output tokens: 128000
LiveCodeBench v6: Overall (Show measurement, test conditions, and source)
- Source value
- 87.313
- Score
- 87.313
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 1
- Category
- coding
- Task
- overall
- Task label
- Overall
- Direction
- Higher is better
- Comparison cohort
- vals-lcb:1:overall
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Source model ID: openai/gpt-5.3-codex. Source model name: GPT-5.3-Codex. Provider reported by source: OpenAI. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-09-01. Reasoning effort: xhigh. Maximum output tokens: 128000. Latency in seconds: 132.631. Standard error: 0.968. Cost per task in USD: 0.091585. Source ID: vals-ai. Evidence capture ID: vals-lcb-2026-10-15-1a7c7dabc9d9. Source payload hash: 1a7c7dabc9d95a0182886b1edc42623987cac2095a0d407a5e334f37acdfa2f5. Parser version: vals-astro-v3. Imported at: 2026-10-15
- Comparison methodology
- Reasoning effort: xhigh. Maximum output tokens: 128000
SWE-bench Verified: <15 min fix (Show measurement, test conditions, and source)
- Source value
- 90.206
- Score
- 90.206
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 1
- Category
- coding
- Task
- <15 min fix
- Task label
- <15 min fix
- Direction
- Higher is better
- Comparison cohort
- vals-swebench:1:<15 min fix
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Source model ID: openai/gpt-5.3-codex. Source model name: GPT-5.3-Codex. Provider reported by source: OpenAI. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-09-01. Reasoning effort: xhigh. Maximum output tokens: 128000. Latency in seconds: 185.93. Standard error: 2.14. Cost per task in USD: 0.340336. Source ID: vals-ai. Evidence capture ID: vals-swebench-2026-10-15-7a4e2e848528. Source payload hash: 7a4e2e8485283caac500127419f68ec4bd5414c837ade114fc6afa5b2d34ccaa. Parser version: vals-astro-v3. Imported at: 2026-10-15
- Comparison methodology
- Reasoning effort: xhigh. Maximum output tokens: 128000
SWE-bench Verified: >4 hours (Show measurement, test conditions, and source)
- Source value
- 33.333
- Score
- 33.333
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 1
- Category
- coding
- Task
- >4 hours
- Task label
- >4 hours
- Direction
- Higher is better
- Comparison cohort
- vals-swebench:1:>4 hours
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Source model ID: openai/gpt-5.3-codex. Source model name: GPT-5.3-Codex. Provider reported by source: OpenAI. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-09-01. Reasoning effort: xhigh. Maximum output tokens: 128000. Latency in seconds: 412.097. Standard error: 33.333. Cost per task in USD: 0.69549. Source ID: vals-ai. Evidence capture ID: vals-swebench-2026-10-15-7a4e2e848528. Source payload hash: 7a4e2e8485283caac500127419f68ec4bd5414c837ade114fc6afa5b2d34ccaa. Parser version: vals-astro-v3. Imported at: 2026-10-15
- Comparison methodology
- Reasoning effort: xhigh. Maximum output tokens: 128000
SWE-bench Verified: 1-4 hours (Show measurement, test conditions, and source)
- Source value
- 54.762
- Score
- 54.762
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 1
- Category
- coding
- Task
- 1-4 hours
- Task label
- 1-4 hours
- Direction
- Higher is better
- Comparison cohort
- vals-swebench:1:1-4 hours
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Source model ID: openai/gpt-5.3-codex. Source model name: GPT-5.3-Codex. Provider reported by source: OpenAI. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-09-01. Reasoning effort: xhigh. Maximum output tokens: 128000. Latency in seconds: 405.865. Standard error: 7.773. Cost per task in USD: 0.818585. Source ID: vals-ai. Evidence capture ID: vals-swebench-2026-10-15-7a4e2e848528. Source payload hash: 7a4e2e8485283caac500127419f68ec4bd5414c837ade114fc6afa5b2d34ccaa. Parser version: vals-astro-v3. Imported at: 2026-10-15
- Comparison methodology
- Reasoning effort: xhigh. Maximum output tokens: 128000
SWE-bench Verified: 15 min - 1 hour (Show measurement, test conditions, and source)
- Source value
- 73.18
- Score
- 73.18
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 1
- Category
- coding
- Task
- 15 min - 1 hour
- Task label
- 15 min - 1 hour
- Direction
- Higher is better
- Comparison cohort
- vals-swebench:1:15 min - 1 hour
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Source model ID: openai/gpt-5.3-codex. Source model name: GPT-5.3-Codex. Provider reported by source: OpenAI. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-09-01. Reasoning effort: xhigh. Maximum output tokens: 128000. Latency in seconds: 264.025. Standard error: 2.748. Cost per task in USD: 0.479692. Source ID: vals-ai. Evidence capture ID: vals-swebench-2026-10-15-7a4e2e848528. Source payload hash: 7a4e2e8485283caac500127419f68ec4bd5414c837ade114fc6afa5b2d34ccaa. Parser version: vals-astro-v3. Imported at: 2026-10-15
- Comparison methodology
- Reasoning effort: xhigh. Maximum output tokens: 128000
SWE-bench Verified: Overall (Show measurement, test conditions, and source)
- Source value
- 78
- Score
- 78
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 1
- Category
- coding
- Task
- overall
- Task label
- Overall
- Direction
- Higher is better
- Comparison cohort
- vals-swebench:1:overall
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-10-15
- Methodology
- Source model ID: openai/gpt-5.3-codex. Source model name: GPT-5.3-Codex. Provider reported by source: OpenAI. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-09-01. Reasoning effort: xhigh. Maximum output tokens: 128000. Latency in seconds: 246.527. Standard error: 1.854. Cost per task in USD: 0.455384. Source ID: vals-ai. Evidence capture ID: vals-swebench-2026-10-15-7a4e2e848528. Source payload hash: 7a4e2e8485283caac500127419f68ec4bd5414c837ade114fc6afa5b2d34ccaa. Parser version: vals-astro-v3. Imported at: 2026-10-15
- Comparison methodology
- Reasoning effort: xhigh. Maximum output tokens: 128000
Page 1 of 3
Our tests
SVG short test
One SVG prompt and fixed output settings. Captures are shown only with 0 reported reasoning tokens or documented lack of reasoning controls. Identical test conditions do not make the models' internal compute identical.
Earlier tests
Pelican on a bicycle
A community classic for free-form SVG drawing.
GPT-5.3-Codex
Reasoning (requested): high
Current attempt
- Run date
- Oct 1, 2026
- API provider
- openrouter
- Requested API model
- openai/gpt-5.3-codex
- API model reported in response
- openai/gpt-5.3-codex
- Reasoning setting
- high
- Pinned endpoint
- OpenAI | openai/gpt-5.3-codex-20260224
- Temperature
- Provider default
- Token limit including reasoning
- 8,192 tokens
- Test protocol
- community-visual-tests-v1
- Technical checks
- Passed, not a quality rating
- Response time
- 47.1 s
Visually reviewed. This individual test does not establish overall model quality.
Prompt and test conditions
Generate an SVG of a pelican riding a bicycle
Token limit including reasoning: 8,192 tokens
Task origin (Simon Willison)Popularity
40.5 / 100
Predecessor and successor
Sources and data date
| Type | Evidence and data date |
|---|---|
| Speed | OpenAI, xhigh effort. Output speed after the first streaming chunk, not total response time. Retrieved September 8, 2026. |
| Price | Base API prices without caching or batch discounts. Higher context tiers and other rates are listed under Costs. |
| Input | Documented modalities. Retrieved September 8, 2026. |
| Output | Documented modalities. Retrieved September 8, 2026. |
| Research status | Research date September 8, 2026. 3 source URLs checked. This documents the inspected sources, not an exhaustive inventory of every publication. |
| Not found in the inspected sources | parameters-billions Original wording OpenAI does not publish parameter counts. |
| Not found in the inspected sources | independent-benchmarks Original wording No independent evaluator row was added in this official-source pass. |
| Additional source | Model metadata (retrieved September 8, 2026) · Introducing GPT-5.3-Codex · Editorial description reviewed October 4, 2026 |
| Additional source | Provider model lifecycle (retrieved October 2, 2026) · OpenAI API deprecations · Editorial description reviewed October 4, 2026 |
| Additional source | METR (retrieved August 26, 2026) |
| Additional source | Vals AI (retrieved October 15, 2026) |
| Additional source | Vals AI (retrieved October 15, 2026) |
| Additional source | Vals AI (retrieved October 15, 2026) |
| Additional source | Vals AI (retrieved October 15, 2026) |
| Additional source | Vals AI (retrieved October 15, 2026) |
| Additional source | OpenRouter: Popularity index (derived), CC BY 4.0 |
| Additional source | Vercel AI Gateway: Popularity index (derived), CC BY 4.0 |
| Additional source | Creative Commons Attribution 4.0 |