gpt-oss-20b
- Input
- $0.07/M tok
- Output
- $0.215/M tok
- Blended
- $0.106/M tok · 3:1
- Speed
- 240tok/s
- Context
- 131Ktokens
- Cost per task
- ~$0.09/taskest.
Long-tail import from Artificial Analysis (2026-09-24); AA variant: gpt-oss-20b (high). Price = AA median host price. Weights ship in MXFP4 (card: "post-trained with MXFP4 quantization of the MoE weights … gpt-oss-20b … within 16GB of memory").
Rankings
Coding
Verified#97 / 137rank by quality34.6±6.5quality / 1003/7 benchmarks measuredWriting & chat
Verified#105 / 162rank by quality30.5±4.1quality / 1003/5 benchmarks measuredHard reasoning
Verified#122 / 218rank by quality37.3±3.3quality / 1004/4 benchmarks measuredPrice frontierAgents
Verified#116 / 187rank by quality23.5±4.3quality / 1004/7 benchmarks measured
Self-hostingEstimated
How we estimate →What it would cost to run these open weights yourself on rented GPUs, compared with the API price above.
- Hardware
- 1× B200 180GB
- BF16 weights
- Throughput
- ~11,917 tok/s
- many requests batched
- Price
- $0.058–0.52 /M tok
- busy → light use
- On your own machine
- RTX 5090 32GB (int4, 300+ tok/s single-stream)
- single consumer GPU or Mac
Assumptions (5)
- BF16 weights (21B params, 3.6B active per token) + 25% KV-cache headroom ≈ 53 GB
- 1× B200 180GB at $5.98–$14.24/GPU-hour on-demand (2026-09-24)
- ~11,917 output tok/s aggregate at batch 256 (bandwidth-bound); MoE compute scales with active params
- Blended 3:1 input:output; prefill ~147,656 tok/s
- Low = 75% utilization at the low GPU price; high = 20% at the high price
Benchmark scores
~ italic, dashed = no published score yet, estimated from related benchmarks.
Coding
SWE-bench VerifiedEstimated
~#17 of 21 (estimated)~69.2%
- #1Claude Opus 596.0%
- #2Claude Sonnet 585.2%
- #16Nemotron 3 Ultra70.7%
- ~#17gpt-oss-20bEstimated69.2%
- #17Intern-S2-Preview (35B-A3B)64.0%
No published score — estimated from related benchmarks
SWE-bench Pro (Public)Estimated
~#23 of 24 (estimated)~47.3%
- #1Claude Fable 5.293.0%
- #2Claude Opus 5.589.9%
- #22Muse Glimmer51.2%
- ~#23gpt-oss-20bEstimated47.3%
- #23Nex-N2.5-Mini43.8%
No published score — estimated from related benchmarks
DeepSWE v1.1Estimated
~#41 of 46 (estimated)~40.7%
- #1Gemini 4 Argon78.8%
- #2DeepSeek V4.1 Flash 101777.0%
- #40Qwen3.8 27B42.2%
- ~#41gpt-oss-20bEstimated40.7%
- #41Laguna S 2.140.4%
No published score — estimated from related benchmarks
Terminal-Bench 4.0Verified
#63 of 1010.0%
- #1Kimi K3.1 Pro67.2%
- #2Claude Sonnet 5.563.6%
- #63gpt-oss-120b0.0%
- #63gpt-oss-20b0.0%
- #63Granite 4.2 30B0.0%
artificialanalysis.ai · 2026-09-24
LiveCodeBenchVerified
#12 of 2777.7%
- #1Qwen3.8-Omni-Flash92.6%
- #2Qwen3.8 Flash-Next91.9%
- #11Gemma 4 31B80.0%
- #12gpt-oss-20b77.7%
- #13Bonsai 27B76.4%
artificialanalysis.ai · 2026-09-24
SciCodeVerified
#85 of 11838.9%
- #1Claude Fable 5.267.7%
- #2Claude Opus 5.566.9%
- #84Intern-S2-Preview (35B-A3B)39.6%
- #85gpt-oss-20b38.9%
- #86Mistral Small 438.8%
artificialanalysis.ai · 2026-09-24
AA Coding Agent IndexEstimated
~#25 of 26 (estimated)~27.9
- #1Claude Opus 5.561.7
- #2Claude Opus 559.7
- #24GPT-6 Luna31.1
- ~#25gpt-oss-20bEstimated27.9
- #25Muse Spark Open26.3
No published score — estimated from related benchmarks
Writing & chat
IFBenchVerified
#51 of 13965.1%
- #1MiniMax-M382.9%
- #2Qwen3.8 2.4T A95B82.8%
- #49Olmo 3.1 32B Think66.0%
- #51gpt-oss-20b65.1%
- #52Qwen3.6 35B A3B64.4%
artificialanalysis.ai · 2026-09-24
MMMLUEstimated
~last of 5 (estimated)below measured range (<81.3%)
No published score — estimated from related benchmarks
LMArena Text, Non-English (Elo)Verified
#51 of 561299 Elo
- #1Gemini 4 ArgonFew votes1512
- #2Claude Opus 5.51497
- #50Molmo2-8BFew votes1305
- #51gpt-oss-20b1299
- #52Olmo 3.1 32B Instruct1294
lmarena.ai · 2026-09-13
LMArena Text (Elo)Verified
#53 of 561317 Elo
- #1Gemini 4 Argon1525
- #2Claude Opus 5.51507
- #52Ring-flash-2.01321
- #53gpt-oss-20b1317
- #54NVIDIA Nemotron 3 Nano 30B A3B1314
lmarena.ai · 2026-09-13
EQ-Bench Creative Writing v3 (Elo)Estimated
~#24 of 26 (estimated)~1333 Elo
- #1GPT-6 Astra2173
- #2Claude Fable 5.12162
- #23Gemma 4 31B1368
- ~#24gpt-oss-20bEstimated1333
- #24gpt-oss-120b961
No published score — estimated from related benchmarks
Hard reasoning
GPQA DiamondVerified
#117 of 19268.8%
- #1GPT-6 Astra96.1%
- #2Gemini 3.8 Flash95.3%
- #116Phi-4-reasoning-plus68.9%
- #117gpt-oss-20b68.8%
- #118K2-V268.1%
artificialanalysis.ai · 2026-09-24
Humanity's Last ExamVerified
#106 of 19811.0%
- #1Claude Fable 5.274.3%
- #2Claude Opus 5.561.4%
- #106Falcon-H1R-7B11.0%
- #106gpt-oss-20b11.0%
- #106Sarvam 105B11.0%
artificialanalysis.ai · 2026-09-24
AIME (latest)Verified
#12 of 3389.3%
- #1GLM-5.3 NVFP499.5%
- #2Beam 1.0 501B A23B97.7%
- #11Xing4.0-29B-A4B90.0%
- #12gpt-oss-20b89.3%
- #13Gemma 4 31B89.2%
artificialanalysis.ai · 2026-09-24
Agents
τ²-bench (Telecom)Verified
#47 of 11960.2%
- #1Step 3.7 Flash98.5%
- #2DeepSeek-V4-Pro-0813-nvfp4-DSpark98.3%
- #46HyperNova 60B 260563.2%
- #47gpt-oss-20b60.2%
- #48Gemma 4 31B59.9%
artificialanalysis.ai · 2026-09-24
OSWorld-VerifiedEstimated
~#12 of 15 (estimated)~59.3%
- #1Kimi K384.8%
- #2Qwen3.8 27B84.3%
- #11Muse Glimmer65.9%
- ~#12gpt-oss-20bEstimated59.3%
- #12Nemotron 3.5 Super VL NVFP452.3%
No published score — estimated from related benchmarks
OSWorld 2.0Estimated
~last of 5 (estimated)below measured range (<47.5%)
No published score — estimated from related benchmarks
BrowseCompEstimated
~#19 of 22 (estimated)~55.3%
- #1Atria Dawn Preview92.5%
- #2GPT-6 Astra91.5%
- #18Nemotron 3.5 Bolt 74B-A9B70.0%
- ~#19gpt-oss-20bEstimated55.3%
- #19Mistral Medium 3.548.6%
No published score — estimated from related benchmarks
GDPval-AA (Elo)Verified
#81 of 91327 Elo
- #1Claude Opus 5.51866
- #2Muse Wonder Full1857
- #80Mistral Small 4349
- #81gpt-oss-20b327
- #82Trinity Large Thinking324
artificialanalysis.ai · 2026-09-24
AutomationBench-AAVerified
#79 of 840.2%
- #1Claude Fable 5.276.4%
- #2Muse Wonder Full71.3%
- #79gpt-oss-120b0.2%
- #79gpt-oss-20b0.2%
- #79K2 Horizon 0.9B0.2%
artificialanalysis.ai · 2026-09-24