OpenAIgpt-oss-120b
gpt-oss-120b is an OpenAI open-weights reasoning model, released 5 Aug 2025. Its best configuration (high reasoning effort) ranks #423 of 785 on the BenchLeader Index at 48.5 ±5.7, in the lower half. The ± is the point: 351 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in instruction following (68) and lowest in knowledge (39). At $0.262 per million tokens blended it is cheaper than most ranked models. Output speed of 208 tokens per second puts it in the fastest quarter, with a first answer in 10.4 s. It has been measured at 4 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 5 Aug 2025.stale: no new result in six months
- Blended price
- $0.262/M
- $0.150 in · $0.600 out
- Output speed
- 208 tok/s
- measured by Artificial Analysis
- First answer
- 10 s
- first token 0.44 s
- Context
- 131k
- Full answer
- 13 s
- median, reasoning included
- Cost per run
- $0.108
- one full Intelligence Index run
- Released
- 5 Aug 2025
- Overall index49
- Reasoning49
- Coding40
- Agents & tools45
- Maths51
- Knowledge39
- Instruction following68
- Long context51
- Composite42
Reasoning-effort configurations
The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Not stated” collects results from publishers that did not say which setting they used; for a reasoning model that is usually its thinking mode, but we do not assume it. Pick a setting here or at the top of the page to see its price, speed and category scores.
| Effort | Index | Rank | Speed | First answer | Chat reply cost | Categories |
|---|---|---|---|---|---|---|
| low | 44.0 | #554 | 261 tok/s | 8.51 s | $0.0002 | Agents & tools 33 · Coding 35 · Composite 41 · Instruction following 59 · Knowledge 37 · Long context 47 · Reasoning 41 |
| medium | – | – | 69 tok/s | 0.44 s | $0.0002 | Coding 44 |
| highbest | 48.5 | #423 | 208 tok/s | 10 s | $0.0002 | Agents & tools 45 · Coding 40 · Composite 42 · Instruction following 68 · Knowledge 39 · Long context 51 · Maths 51 · Reasoning 49 |
| not stated | 45.6 | #508 | 69 tok/s | 0.44 s | $0.0002 | Agents & tools 23 · Coding 43 · Human preference 52 · Instruction following 47 · Knowledge 47 · Maths 64 · Reasoning 41 |
How its index has moved
8 Sept 2026 to 23 Oct 2026- high
- not stated
- low
The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.
Benchmark results
See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | low | medium | high | not stated | Source | Trend |
|---|---|---|---|---|---|---|
| GPQA Diamond | – | – | 75.8%#144 | – | Epoch AI Benchmarking Hub | |
| SimpleBencheffort not stated | – | – | – | 22.1%#80 | SimpleBench | |
| Arena Hard Promptseffort not stated | – | – | – | 1362#205 | Arena (LMArena) | |
| GPQA Diamond (AA)not in index | 67.2%#322 | – | 78.2%#212 | – | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 5.9%#413 | – | 19.6%#218 | – | Artificial Analysis | |
| GPQA Diamond (Vals)not in indexeffort not stated | – | – | – | 78.5%#70 | Vals AI | |
| Kagi LLM Benchmarkeffort not stated | – | – | – | 58.6%#53 | Kagi LLM Benchmark | |
| GPQA Diamond (HELM)not in indexeffort not stated | – | – | – | 68.4%#12 | HELM Capabilities | |
| CritPt | 0.0%#315 | – | 1.1%#223 | – | Artificial Analysis | |
| Mystery Game Puzzlesnot in index | – | 2.0%#140 | 0.0%#141 | – | Epoch AI Benchmarking Hub | |
| LMCAnot in index | – | – | 22.1%#156 | – | Epoch AI Benchmarking Hub | |
| DTBenchnot in index | – | – | 76.3%#146 | – | Epoch AI Benchmarking Hub |
Coding
| Benchmark | low | medium | high | not stated | Source | Trend |
|---|---|---|---|---|---|---|
| SciCode | 36.0%#186 | – | 34.0%#193 | – | SciCode | |
| WeirdML | – | 41.9%#95 | 48.2%#73 | – | WeirdML | |
| Arena Codingeffort not stated | – | – | – | 1389#203 | Arena (LMArena) | |
| SciCode (AA)not in index | – | – | 34.0%#202 | – | Artificial Analysis | |
| LiveCodeBencheffort not stated | – | – | – | 83.2%#49 | Vals AI | |
| SWE-bench (Vals)not in indexeffort not stated | – | – | – | 33.6%#84 | Vals AI | |
| SWE-Bench Proeffort not stated | – | – | – | 16.2%#19 | Scale AI SEAL | |
| Aider Polygloteffort not stated | – | – | – | 41.8%#30 | Aider polyglot leaderboard | |
| SWE-bench Verified (bash only)effort not stated | – | – | – | 26.0%#39 | SWE-bench | |
| SWE-bench Verified (any scaffold)not in indexeffort not stated | – | – | – | 26.0%#53 | SWE-bench | |
| SWE-Bench Pro (private)not in indexeffort not stated | – | – | – | 16.2%#19 | Scale AI SEAL | |
| Surface Evolver Benchnot in indexeffort not stated | – | – | – | 25.0%#27 | Epoch AI Benchmarking Hub | |
| ALE-Benchnot in indexeffort not stated | – | – | – | 575.6#100 | Epoch AI Benchmarking Hub | |
| AlgoTunenot in index | – | – | 1.4#14 | – | AlgoTune | |
| Terminal-Bench 4.0 (AA)not in index | – | – | 0.0%#152 | – | Artificial Analysis | |
| Terminal-Bench 2.1 (AA)not in index | 13.9%#184 | – | 26.2%#165 | – | Artificial Analysis |
Agents & tools
| Benchmark | low | medium | high | not stated | Source | Trend |
|---|---|---|---|---|---|---|
| Terminal-Bencheffort not stated | – | – | – | 18.7%#73 | Terminal-Bench | |
| APEX-Agentseffort not stated | – | – | – | 4.4%#53 | Mercor | |
| Terminal-Bench Hard | 5.3%#277 | – | 23.5%#150 | – | Artificial Analysis | |
| τ²-Bench Telecom (AA)not in index | 45.0%#212 | – | 65.8%#165 | – | Artificial Analysis | |
| GDPval-AA v2.1not in index | 0.0%#124 | – | 5.4%#116 | – | Artificial Analysis | |
| τ³-Banking (AA) | 2.9%#199 | – | 12.8%#130 | – | Artificial Analysis | |
| ITBench SRE (AA)not in index | – | – | 5.7%#43 | – | Artificial Analysis | |
| APEX-Agents (AA)not in index | – | – | 3.1%#30 | – | Artificial Analysis | |
| Poker Agentnot in indexeffort not stated | – | – | – | 1015.3%#12 | Vals AI | |
| Terminal-Bench 2.0 (Vals)not in indexeffort not stated | – | – | – | 19.1%#56 | Vals AI | |
| METR Time Horizonsnot in indexeffort not stated | – | – | – | 56.6%#22 | METR | |
| Vending-Bench 2not in indexeffort not stated | – | – | – | -21.5#67 | Andon Labs | |
| AutomationBenchnot in index | – | – | 0.2%#217 | – | Artificial Analysis | |
| GDP.pdf (AA)not in index | – | – | 4.0%#170 | – | Artificial Analysis | |
| EnterpriseOps-Gymnot in index | – | – | 25.5%#47 | – | Artificial Analysis | |
| Terminal-Bench-Science 0.1 (AA)not in index | – | – | 0.0%#71 | – | Artificial Analysis | |
| BrowseComp-Plusnot in index | 9.8%#17 | 24.6%#7 | 43.6%#3 | – | BrowseComp-Plus | |
| BrowseCompnot in indexeffort not stated | – | – | – | 1.6%#11 | OpenAI |
Maths
| Benchmark | high | not stated | Source | Trend |
|---|---|---|---|---|
| OTIS Mock AIME | 88.9%#76 | – | Epoch AI Benchmarking Hub | |
| AIME (Vals)effort not stated | – | 92.6%#18 | Vals AI | |
| MGSMeffort not stated | – | 92.0%#24 | Vals AI | |
| MathArena Apex | 1.0%#35 | – | MathArena | |
| Omni-MATH (HELM)effort not stated | – | 68.8%#5 | HELM Capabilities | |
| MATH 500not in indexeffort not stated | – | 94.8%#6 | Vals AI | |
| Arena Mathseffort not stated | – | 1381#171 | Arena (LMArena) |
Knowledge
| Benchmark | low | high | not stated | Source | Trend |
|---|---|---|---|---|---|
| AA-Omniscience | -53.5#421 | -49.3#384 | – | Artificial Analysis | |
| MMLU-Proeffort not stated | – | – | 79.2%#93 | Vals AI | |
| LegalBencheffort not stated | – | – | 75.9%#108 | Vals AI | |
| CorpFineffort not stated | – | – | 58.2%#82 | Vals AI | |
| TaxEvaleffort not stated | – | – | 71.6%#73 | Vals AI | |
| MedQAeffort not stated | – | – | 91.4%#38 | Vals AI | |
| PRBench Financeeffort not stated | – | – | 43.8%#22 | Scale AI SEAL | |
| PRBench Legaleffort not stated | – | – | 40.2%#32 | Scale AI SEAL | |
| MultiNRCeffort not statedpossible contamination | – | – | 15.2%#39 | Scale AI SEAL | |
| MMLU-Pro (HELM)not in indexeffort not stated | – | – | 79.5%#20 | HELM Capabilities | |
| CaseLaw v2not in indexeffort not stated | – | – | 48.8%#49 | Vals AI | |
| AA-Omniscience: accuracynot in index | 19.8%#309 | 21.8%#281 | – | Artificial Analysis | |
| AA-Omniscience: non-hallucinationnot in index | 8.6%#453 | 9.2%#434 | – | Artificial Analysis |
Instruction following
| Benchmark | low | high | not stated | Source | Trend |
|---|---|---|---|---|---|
| IFBench | 58.3%#132 | 69.0%#76 | – | Artificial Analysis | |
| MultiChallengeeffort not stated | – | – | 45.3%#28 | Scale AI SEAL | |
| IFEval (HELM)effort not stated | – | – | 83.6%#30 | HELM Capabilities | |
| WildBench (HELM)effort not stated | – | – | 84.5%#14 | HELM Capabilities | |
| Arena Instruction Followingnot in indexeffort not stated | – | – | 1324#213 | Arena (LMArena) |
Human preference
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Arena Text | 1352#200 | Arena (LMArena) | |
| Arena Creative Writingnot in index | 1279#238 | Arena (LMArena) | |
| Arena Multi-turnnot in index | 1327#215 | Arena (LMArena) | |
| Arena Longer Queriesnot in index | 1324#237 | Arena (LMArena) | |
| Lech Mazur Creative Writingnot in index | 77.1%#20 | Lech Mazur |
Long context
| Benchmark | low | high | Source | Trend |
|---|---|---|---|---|
| AA-LCR | 46.0%#333 | 52.0%#317 | Artificial Analysis | |
| MLCRnot in index | – | 1.7%#162 | Artificial Analysis |
Composite
| Benchmark | low | high | not stated | Source | Trend |
|---|---|---|---|---|---|
| Epoch Capabilities Indexnot in indexeffort not stated | – | – | 139.8#113 | Epoch AI Benchmarking Hub | |
| AA Intelligence Index v4.3.2 | 10.2#381 | 11.6#349 | – | Artificial Analysis | |
| HELM Capabilities meannot in indexeffort not stated | – | – | 77.0%#9 | HELM Capabilities |
Safety & honesty
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| FORTRESSnot in index | 8.2%#67 | Scale AI SEAL | |
| MASKnot in indexpossible contamination | 92.0%#6 | Scale AI SEAL |
Where to run it
Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| Cerebras | 805 tok/s | 0.20 s | $0.350 | $0.750 | 131k | fp16 |
| SambaNova | 398 tok/s | 1.18 s | $0.140 | $0.950 | 131k | – |
| Groq | 344 tok/s | 0.23 s | $0.150 | $0.600 | 131k | – |
| Nebius Token Factory | 314 tok/s | 0.29 s | $0.150 | $0.600 | 131k | fp4 |
| Crusoe | 261 tok/s | 0.19 s | $0.050 | $0.250 | 131k | bf16 |
| Amazon Bedrock | 208 tok/s | 0.41 s | $0.150 | $0.600 | 131k | – |
| MARA | 196 tok/s | 0.98 s | $0.150 | $0.750 | 131k | – |
| Together | 192 tok/s | 0.21 s | $0.150 | $0.600 | 131k | – |
| Phala | 188 tok/s | 0.81 s | $0.150 | $0.600 | 131k | – |
| Parasail | 169 tok/s | 0.45 s | $0.100 | $0.750 | 131k | fp4 |
| NovitaAI | 145 tok/s | 0.76 s | $0.050 | $0.250 | 131k | fp4 |
| Baseten | 138 tok/s | 0.18 s | $0.100 | $0.500 | 128k | fp4 |
| Baseten | 138 tok/s | 0.19 s | $0.100 | $0.500 | 128k | fp4 |
| Amazon Bedrock | 120 tok/s | 0.62 s | $0.150 | $0.600 | 131k | – |
| AkashML | 110 tok/s | 0.35 s | $0.030 | $0.187 | 131k | bf16 |
| CoreWeave | 107 tok/s | 0.38 s | $0.030 | $0.170 | 131k | fp4 |
| SiliconFlow | 78 tok/s | 1.16 s | $0.150 | $0.600 | 131k | fp8 |
| DekaLLM | 68 tok/s | 0.59 s | $0.030 | $0.180 | 131k | bf16 |
| Mancer | 66 tok/s | 0.45 s | $0.050 | $0.250 | 131k | fp8 |
| DigitalOcean | 60 tok/s | 0.43 s | $0.060 | $0.420 | 128k | – |
| DeepInfra | 59 tok/s | 0.31 s | $0.037 | $0.170 | 131k | bf16 |
| NEAR AI | 47 tok/s | 0.58 s | $0.030 | $0.150 | 131k | fp4 |
| Venice | 30 tok/s | 1.75 s | $0.030 | $0.150 | 128k | int4 |
| Google Vertex | 28 tok/s | 1.10 s | $0.090 | $0.360 | 131k | – |
Price history
The price above, per 1M tokens, every day it changed.
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $0.100 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0002 | $0.0002 | 11.9 s |
| Summarize a 30-page report | 12,000 / 600 | $0.0022 | $0.0017 | 13.3 s |
| Code edit | 6,000 / 1,500 | $0.0018 | $0.0016 | 17.7 s |
| Agentic coding session | 60,000 / 4,000 | $0.011 | $0.0092 | 29.7 s |
| Structured extraction | 2,000 / 200 | $0.0004 | $0.0003 | 11.4 s |
More from OpenAI
See also
Questions people ask
- What are gpt-oss-120b's benchmark scores?
- As of 24 Oct 2026, gpt-oss-120b (high effort) scores 48.5 ±5.7 on the BenchLeader Index, strongest in instruction following (68) and weakest in knowledge (39). That rests on 12 independent benchmark results, and every score on this page links to its publisher.
- How does gpt-oss-120b rank?
- gpt-oss-120b (high effort) is #423 of 785 ranked configurations on 24 Oct 2026. 351 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 72.6.
- How much does gpt-oss-120b cost per 1M tokens?
- As of 24 Oct 2026, gpt-oss-120b costs $0.150 per 1M input tokens and $0.600 per 1M output tokens, $0.262 blended (three input tokens per output token). The list price last changed on 21 Oct 2026.
- When was gpt-oss-120b released?
- OpenAI released gpt-oss-120b on 5 Aug 2025. BenchLeader has results for it at 4 reasoning-effort settings.
Data as of 24 Oct 2026. Compare these configurations.
Cite as: BenchLeader, “gpt-oss-120b: benchmarks, pricing, speed and rank”, data as of 24 Oct 2026, https://www.benchleader.com/models/gpt-oss-120b.