GPT OSS 20B
Not tested yet
Language modelActive
GPT OSS 20B
OpenAI
- Released
- August 5, 2025
- Data date
- September 8, 2026
- Intelligence
- Overall score
- Speed
- 189.1 tokens/sOutput throughput
- Price
- Input · Output / 1M tokens
- Input
- Text
- Output
- Text
No published category-wide ranking is available for this model.
Entries 1-12 of 79. Page 1 of 7.
Published benchmarks
Each chart compares models in the same documented test and highlights the current model.
Finance
2 measurementsCorpFin v2
Overall
Higher is better50 modelsGPT OSS 20B: Rank 45 of 50Data date: September 4, 2026
Scale: 0% - 100%
Methodology and context
- Comparison cohort
- vals-corp-fin-v2:2:overall
- Metric
- accuracy
- Scale
- 0-100
- Participants
- 50
- Methodology and context
- Source model ID: fireworks/gpt-oss-20b. Source model name: GPT OSS 20B. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-08-12. Latency in seconds: 33.968. Cost per task in USD: 0.004549. Source ID: vals-ai. Evidence capture ID: vals-corp_fin_v2-2026-09-04-46dad6684551. Source payload hash: 46dad668455190ee4239b7c743d00d81c6fd3127e0d87123c518ec107c13b44d. Parser version: vals-astro-v2. Imported at: 2026-09-04
TaxEval v2
Overall
Higher is better54 modelsGPT OSS 20B: Rank 52 of 54Data date: September 4, 2026
Scale: 0% - 100%
Methodology and context
- Comparison cohort
- vals-tax-eval-v2:2:overall
- Metric
- accuracy
- Scale
- 0-100
- Participants
- 54
- Methodology and context
- Source model ID: fireworks/gpt-oss-20b. Source model name: GPT OSS 20B. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-09-01. Latency in seconds: 49.004. Cost per task in USD: 0.002254. Source ID: vals-ai. Evidence capture ID: vals-tax_eval_v2-2026-09-04-f0e140ab16fc. Source payload hash: f0e140ab16fcb52041f4f87d2872273c03ca8014fdb69b3da9afd01b0f1e77ec. Parser version: vals-astro-v2. Imported at: 2026-09-04
GPQA Diamond
Overall
Higher is better54 modelsGPT OSS 20B: Rank 46 of 54Data date: September 4, 2026
Scale: 0% - 100%
Methodology and context
- Comparison cohort
- vals-gpqa:1:overall
- Metric
- accuracy
- Scale
- 0-100
- Participants
- 54
- Methodology and context
- Source model ID: fireworks/gpt-oss-20b. Source model name: GPT OSS 20B. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-09-01. Latency in seconds: 145.092. Cost per task in USD: 0.006738. Source ID: vals-ai. Evidence capture ID: vals-gpqa-2026-09-04-fc438726f0d8. Source payload hash: fc438726f0d8523d320f02d75ae372a3fa38b28b36f70fe318d119f55153560d. Parser version: vals-astro-v2. Imported at: 2026-09-04
LiveCodeBench v6
Overall
Higher is better56 modelsGPT OSS 20B: Rank 40 of 56Data date: September 4, 2026
Scale: 0% - 100%
Methodology and context
- Comparison cohort
- vals-lcb:1:overall
- Metric
- accuracy
- Scale
- 0-100
- Participants
- 56
- Methodology and context
- Source model ID: fireworks/gpt-oss-20b. Source model name: GPT OSS 20B. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-09-01. Latency in seconds: 109.453. Cost per task in USD: 0.006706. Source ID: vals-ai. Evidence capture ID: vals-lcb-2026-09-04-95ca83387127. Source payload hash: 95ca83387127df287e9a6d928c411c98c9b65b4cefeee46eb4008ca1521f5ed4. Parser version: vals-astro-v2. Imported at: 2026-09-04
LegalBench
Overall
Higher is better53 modelsGPT OSS 20B: Rank 48 of 53Data date: September 4, 2026
Scale: 0% - 100%
Methodology and context
- Comparison cohort
- vals-legal-bench:1:overall
- Metric
- accuracy
- Scale
- 0-100
- Participants
- 53
- Methodology and context
- Source model ID: fireworks/gpt-oss-20b. Source model name: GPT OSS 20B. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-09-01. Latency in seconds: 7.187. Source ID: vals-ai. Evidence capture ID: vals-legal_bench-2026-09-04-ff81649923fb. Source payload hash: ff81649923fb684ff37d5f557fd320ba5ab433927831065ed781a94548f480ef. Parser version: vals-astro-v2. Imported at: 2026-09-04
Mathematics
2 measurementsMATH 500
Overall
Higher is better17 modelsGPT OSS 20B: Rank 5 of 17Data date: September 4, 2026
Scale: 0% - 100%
Methodology and context
- Comparison cohort
- vals-math500:1:overall
- Metric
- accuracy
- Scale
- 0-100
- Participants
- 17
- Methodology and context
- Source model ID: fireworks/gpt-oss-20b. Source model name: GPT OSS 20B. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-01-09. Latency in seconds: 31.472. Cost per task in USD: 0.000955. Source ID: vals-ai. Evidence capture ID: vals-math500-2026-09-04-a87fdb25617d. Source payload hash: a87fdb25617d5588f5c63d5774177861d27be3a848b9eb67a7f417d2b97cb127. Parser version: vals-astro-v2. Imported at: 2026-09-04
MGSM
Overall
Higher is better21 modelsGPT OSS 20B: Rank 16 of 21Data date: September 4, 2026
Scale: 0% - 100%
Methodology and context
- Comparison cohort
- vals-mgsm:1:overall
- Metric
- accuracy
- Scale
- 0-100
- Participants
- 21
- Methodology and context
- Source model ID: fireworks/gpt-oss-20b. Source model name: GPT OSS 20B. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-01-09. Latency in seconds: 5.791. Cost per task in USD: 0.000453. Source ID: vals-ai. Evidence capture ID: vals-mgsm-2026-09-04-8454c4c2a5d6. Source payload hash: 8454c4c2a5d6a567e4c17ce67f5f6eb9f3589353348190378d4b83b2df182750. Parser version: vals-astro-v2. Imported at: 2026-09-04
MedQA
Overall
Higher is better27 modelsGPT OSS 20B: Rank 22 of 27Data date: September 4, 2026
Scale: 0% - 100%
Methodology and context
- Comparison cohort
- vals-medqa:1:overall
- Metric
- accuracy
- Scale
- 0-100
- Participants
- 27
- Methodology and context
- Source model ID: fireworks/gpt-oss-20b. Source model name: GPT OSS 20B. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-04-16. Latency in seconds: 25.351. Cost per task in USD: 0.000335. Source ID: vals-ai. Evidence capture ID: vals-medqa-2026-09-04-de083a524dbf. Source payload hash: de083a524dbf0fa1c2da5cf0a06bd001693d2426690681ffac4cd93ccf3fbeba. Parser version: vals-astro-v2. Imported at: 2026-09-04
MMLU-Pro
Overall
Higher is better55 modelsGPT OSS 20B: Rank 52 of 55Data date: September 4, 2026
Scale: 0% - 100%
Methodology and context
- Comparison cohort
- vals-mmlu-pro:1:overall
- Metric
- accuracy
- Scale
- 0-100
- Participants
- 55
- Methodology and context
- Source model ID: fireworks/gpt-oss-20b. Source model name: GPT OSS 20B. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-09-01. Latency in seconds: 30.207. Cost per task in USD: 0.00062. Source ID: vals-ai. Evidence capture ID: vals-mmlu_pro-2026-09-04-94eef672a38c. Source payload hash: 94eef672a38c8146fc1334fecbd173c083343c88de0f3047ae7d034776227e0f. Parser version: vals-astro-v2. Imported at: 2026-09-04
Measurements without matching peer values
These measurements have no matching peer values under the same test conditions. Their original values and sources remain available here.
AIME: Overall (Show measurement, test conditions, and source)
- Source value
- 86.042
- Score
- 86.042
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 1
- Category
- math
- Task
- overall
- Task label
- Overall
- Direction
- Higher is better
- Comparison cohort
- vals-aime:1:overall
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-09-04
- Methodology
- Source model ID: fireworks/gpt-oss-20b. Source model name: GPT OSS 20B. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-04-16. Latency in seconds: 244.712. Cost per task in USD: 0.011043. Source ID: vals-ai. Evidence capture ID: vals-aime-2026-09-04-ab904d5655de. Source payload hash: ab904d5655de635c6d89d8e7b15a28996af0ceb1900d3589b2676c3de5eb9ab7. Parser version: vals-astro-v2. Imported at: 2026-09-04
- Context
- The source dataset marks this benchmark as archived. The result remains visible as historical evidence and is not interchangeable with an active leaderboard.
CaseLaw v2: Overall (Show measurement, test conditions, and source)
- Source value
- 43.837
- Score
- 43.837
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 2
- Category
- legal
- Task
- overall
- Task label
- Overall
- Direction
- Higher is better
- Comparison cohort
- vals-case-law-v2:2:overall
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-09-04
- Methodology
- Source model ID: fireworks/gpt-oss-20b. Source model name: GPT OSS 20B. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-05-04. Latency in seconds: 85.696. Cost per task in USD: 0.002158. Source ID: vals-ai. Evidence capture ID: vals-case_law_v2-2026-09-04-4e6c3c195dfb. Source payload hash: 4e6c3c195dfb3be9c003687ced5c27ccfe375f6137c4145a318f033617209537. Parser version: vals-astro-v2. Imported at: 2026-09-04
- Context
- The source dataset marks this benchmark as archived. The result remains visible as historical evidence and is not interchangeable with an active leaderboard.
AIME 2024 (Show measurement, test conditions, and source)
- Source value
- 42.1
- Score
- 42.1
- Metric
- percent
- Unit
- %
- Category
- Academic
- Direction
- Higher is better
- Comparison cohort
- research:gpt-oss-20b:0
- Source type
- vendor-reported
- Evaluator
- OpenAI
- Status
- active
- Retrieved at
- 2026-09-08
- Methodology
- Vendor-reported result. gpt-oss-20b; low reasoning; no tools
- Context
- Official model-card percentage; tool/no-tool setup is retained in configuration.
Recorded measurement details
Every observation retains its source value and published test conditions.
CorpFin v2: Overall (Show measurement, test conditions, and source)
- Source value
- 53.147
- Score
- 53.147
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 2
- Category
- finance
- Task
- overall
- Task label
- Overall
- Direction
- Higher is better
- Comparison cohort
- vals-corp-fin-v2:2:overall
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-09-04
- Methodology
- Source model ID: fireworks/gpt-oss-20b. Source model name: GPT OSS 20B. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-08-12. Latency in seconds: 33.968. Cost per task in USD: 0.004549. Source ID: vals-ai. Evidence capture ID: vals-corp_fin_v2-2026-09-04-46dad6684551. Source payload hash: 46dad668455190ee4239b7c743d00d81c6fd3127e0d87123c518ec107c13b44d. Parser version: vals-astro-v2. Imported at: 2026-09-04
GPQA Diamond: Overall (Show measurement, test conditions, and source)
- Source value
- 68.94
- Score
- 68.94
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 1
- Category
- reasoning
- Task
- overall
- Task label
- Overall
- Direction
- Higher is better
- Comparison cohort
- vals-gpqa:1:overall
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-09-04
- Methodology
- Source model ID: fireworks/gpt-oss-20b. Source model name: GPT OSS 20B. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-09-01. Latency in seconds: 145.092. Cost per task in USD: 0.006738. Source ID: vals-ai. Evidence capture ID: vals-gpqa-2026-09-04-fc438726f0d8. Source payload hash: fc438726f0d8523d320f02d75ae372a3fa38b28b36f70fe318d119f55153560d. Parser version: vals-astro-v2. Imported at: 2026-09-04
LiveCodeBench v6: Overall (Show measurement, test conditions, and source)
- Source value
- 80.387
- Score
- 80.387
- 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
- 2026-09-04
- Methodology
- Source model ID: fireworks/gpt-oss-20b. Source model name: GPT OSS 20B. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-09-01. Latency in seconds: 109.453. Cost per task in USD: 0.006706. Source ID: vals-ai. Evidence capture ID: vals-lcb-2026-09-04-95ca83387127. Source payload hash: 95ca83387127df287e9a6d928c411c98c9b65b4cefeee46eb4008ca1521f5ed4. Parser version: vals-astro-v2. Imported at: 2026-09-04
LegalBench: Overall (Show measurement, test conditions, and source)
- Source value
- 70.849
- Score
- 70.849
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 1
- Category
- legal
- Task
- overall
- Task label
- Overall
- Direction
- Higher is better
- Comparison cohort
- vals-legal-bench:1:overall
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-09-04
- Methodology
- Source model ID: fireworks/gpt-oss-20b. Source model name: GPT OSS 20B. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-09-01. Latency in seconds: 7.187. Source ID: vals-ai. Evidence capture ID: vals-legal_bench-2026-09-04-ff81649923fb. Source payload hash: ff81649923fb684ff37d5f557fd320ba5ab433927831065ed781a94548f480ef. Parser version: vals-astro-v2. Imported at: 2026-09-04
MATH 500: Overall (Show measurement, test conditions, and source)
- Source value
- 94.2
- Score
- 94.2
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 1
- Category
- math
- Task
- overall
- Task label
- Overall
- Direction
- Higher is better
- Comparison cohort
- vals-math500:1:overall
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-09-04
- Methodology
- Source model ID: fireworks/gpt-oss-20b. Source model name: GPT OSS 20B. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-01-09. Latency in seconds: 31.472. Cost per task in USD: 0.000955. Source ID: vals-ai. Evidence capture ID: vals-math500-2026-09-04-a87fdb25617d. Source payload hash: a87fdb25617d5588f5c63d5774177861d27be3a848b9eb67a7f417d2b97cb127. Parser version: vals-astro-v2. Imported at: 2026-09-04
MedQA: Overall (Show measurement, test conditions, and source)
- Source value
- 82.875
- Score
- 82.875
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 1
- Category
- health
- Task
- overall
- Task label
- Overall
- Direction
- Higher is better
- Comparison cohort
- vals-medqa:1:overall
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-09-04
- Methodology
- Source model ID: fireworks/gpt-oss-20b. Source model name: GPT OSS 20B. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-04-16. Latency in seconds: 25.351. Cost per task in USD: 0.000335. Source ID: vals-ai. Evidence capture ID: vals-medqa-2026-09-04-de083a524dbf. Source payload hash: de083a524dbf0fa1c2da5cf0a06bd001693d2426690681ffac4cd93ccf3fbeba. Parser version: vals-astro-v2. Imported at: 2026-09-04
MGSM: Overall (Show measurement, test conditions, and source)
- Source value
- 89.018
- Score
- 89.018
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 1
- Category
- math
- Task
- overall
- Task label
- Overall
- Direction
- Higher is better
- Comparison cohort
- vals-mgsm:1:overall
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-09-04
- Methodology
- Source model ID: fireworks/gpt-oss-20b. Source model name: GPT OSS 20B. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-01-09. Latency in seconds: 5.791. Cost per task in USD: 0.000453. Source ID: vals-ai. Evidence capture ID: vals-mgsm-2026-09-04-8454c4c2a5d6. Source payload hash: 8454c4c2a5d6a567e4c17ce67f5f6eb9f3589353348190378d4b83b2df182750. Parser version: vals-astro-v2. Imported at: 2026-09-04
MMLU-Pro: Overall (Show measurement, test conditions, and source)
- Source value
- 71.636
- Score
- 71.636
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 1
- Category
- knowledge
- Task
- overall
- Task label
- Overall
- Direction
- Higher is better
- Comparison cohort
- vals-mmlu-pro:1:overall
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-09-04
- Methodology
- Source model ID: fireworks/gpt-oss-20b. Source model name: GPT OSS 20B. Evaluator: Vals AI. Dataset type: public. Leaderboard date: 2026-09-01. Latency in seconds: 30.207. Cost per task in USD: 0.00062. Source ID: vals-ai. Evidence capture ID: vals-mmlu_pro-2026-09-04-94eef672a38c. Source payload hash: 94eef672a38c8146fc1334fecbd173c083343c88de0f3047ae7d034776227e0f. Parser version: vals-astro-v2. Imported at: 2026-09-04
TaxEval v2: Overall (Show measurement, test conditions, and source)
- Source value
- 63.696
- Score
- 63.696
- Metric
- accuracy
- Unit
- %
- Benchmark version
- 2
- Category
- finance
- Task
- overall
- Task label
- Overall
- Direction
- Higher is better
- Comparison cohort
- vals-tax-eval-v2:2:overall
- Source type
- independent-evaluator
- Evaluator
- Vals AI
- Status
- active
- Retrieved at
- 2026-09-04
- Methodology
- Source model ID: fireworks/gpt-oss-20b. Source model name: GPT OSS 20B. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-09-01. Latency in seconds: 49.004. Cost per task in USD: 0.002254. Source ID: vals-ai. Evidence capture ID: vals-tax_eval_v2-2026-09-04-f0e140ab16fc. Source payload hash: f0e140ab16fcb52041f4f87d2872273c03ca8014fdb69b3da9afd01b0f1e77ec. Parser version: vals-astro-v2. Imported at: 2026-09-04
Our model tests
Community tests you can inspect
Compare how the models respond to the same prompt. Each test shows the first attempt, with no subsequent fixes to the generated code. These results do not contribute to an overall score.
Pelican on a bicycle
A community classic for free-form SVG drawing.
Prompt and test conditions
Generate an SVG of a pelican riding a bicycle
Token limit including reasoning: 8,192 tokens
Task origin (Simon Willison)Profile
Specifications and access
Published information about this model. Existing estimates are explicitly labeled.
Additional source observations
Dated documentation and model-card observations. Provider limits, native context and extended context can differ. Configuration notes retain the source wording.
- License
- Apache-2.0Source
- Open weights
- YesSource
- Architecture
- Mixture-of-experts transformerSource
- Parameters
- 21 BSource
- Active parameters
- 3.6 BSource
- Layers
- 24Source
- Total experts
- 32Source
- Active experts per token
- 4Source
- Rounded context label
- 128kSource
- Input
- TextSource
- Output
- TextSource
- Reasoning mode
- low, medium, highSource
- Tool calling
- YesSource
- Structured output
- YesSource
- Knowledge cutoff
- 2024-06Source
- Training tokens
- trillionsSource
- Release
- 2025-08-05Source
Pricing
Published prices
Prices apply to the stated unit. Resolution, output length, and provider can change the cost.
Cost example
100 API requests
$0.0095
100 requests with 1,000 input and 500 output tokens each calculate to $0.0095. Input accounts for $0.003; output accounts for $0.0065.
- Input price
- $0.03 / 1M tokensBase tier for each request
- Output price
- $0.13 / 1M tokensBase tier for each request
The calculation uses documented token prices and no cache discount. It excludes extra tools, tax, reasoning tokens, and further hidden output tokens.
Open sourceHead-to-head comparisons
Compare this model
Each matchup compares this model with exactly one other model from the same category.
There are no published direct comparisons for this model yet.
Evidence
Sources and data date
Every statement links to its underlying documentation or leaderboard.
- Speed
Median across serving providers, High 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 date September 8, 2026. 3 source URLs checked. This documents the inspected sources, not an exhaustive inventory of every publication.
Open evidence questions and research notes (original wording)
Not found in the inspected sources
exact-training-token-count. The model card says trillions of tokens but does not publish an exact count.
Not found in the inspected sources
independent-benchmarks. This artifact retains the complete official model-card benchmark rows used here; no external leaderboard was merged.
- Model metadataSource
- Context window dataSource
- Knowledge cutoff data (retrieved September 8, 2026)Source
- API pricingSource
- Vals AI (retrieved September 4, 2026)Source
- Vals AI (retrieved September 4, 2026)Source
- Vals AI (retrieved September 4, 2026)Source
- Vals AI (retrieved September 4, 2026)Source
- Vals AI (retrieved September 4, 2026)Source
- Vals AI (retrieved September 4, 2026)Source
- Vals AI (retrieved September 4, 2026)Source
- Vals AI (retrieved September 4, 2026)Source
- Vals AI (retrieved September 4, 2026)Source
- Vals AI (retrieved September 4, 2026)Source
- Vals AI (retrieved September 4, 2026)Source
- Artificial Analysis (retrieved September 8, 2026)Source
- Model metadata (retrieved September 8, 2026)Source
- Speed measurement methodologySource