gpt-oss-120b
OpenAIReleased Aug 5, 2025Open weights · Apache 2.0gpt-oss-120b
- Intelligence (Epoch)
- 140 · 80th of 111
- Reasoning
- yes, with effort levels
- Knowledge cutoff
- June 2024
- Context
- 131K tokens
- Max output
- 64k tokens
- Input
- text
- API features
- tools, structured outputs, JSON mode, seed, logprobs
Read Oct 23, 06:00 UTC from the hosts' published offers and OpenRouter, spec sheet from OpenAI · cutoff: platform.openai.com
Intelligence for the price
1 model scores higher for the same price or less; it ranks 2 of the 7 models at its price or below.
Epoch Capabilities Index against the lowest input + output price per M, log scale · prices Oct 23. Score: Epoch AI, CC BY 4.0, 140 with an interval of 136 to 142.
Against its rivals
The best of its price range and the top score
- gpt-oss-120bthis model$0.18140
- Claude Opus 5.5top score$24.00169
- DeepSeek V4 Flash 0731same price range$0.21155
- GLM-5.3-Flashsame price range$0.19153
Bars: Epoch Capabilities Index (Epoch AI, CC BY 4.0), drawn from 100, below the lowest score in the catalogue · price: lowest input + output per M.
Speed by provider
Output speed of gpt-oss-120b at each host, fastest first (the 10 fastest of 18)
- Cerebras932 tok/s0.15 s to first token
- SambaNova388 tok/s1.21 s to first token
- Crusoe352 tok/s0.20 s to first token
- Groq343 tok/s0.16 s to first token
- Nebius305 tok/s0.29 s to first token
- Phala202 tok/s0.84 s to first token
- Parasail184 tok/s0.36 s to first token
- Amazon Bedrock160 tok/s0.43 s to first token
- BaseTen146 tok/s0.17 s to first token
- AkashML134 tok/s0.38 s to first token
Speed: OpenRouter, median of the last 30 min, read Oct 23, 06:00 UTC · a snapshot of real traffic before the daily read, not a daily average · bars from zero.
Run it yourself
open weights, Apache 2.0 · fits 4 of the 6 workstations in our guideSelf-host guide ›- Parameters
- 116.8B
- 5.1B active per token (MoE)
- Memory needed
- 70 GB
- in INT4 · 140 GB in INT8
- Smallest cloud setup
- A100
- 80 GB · $1.19/h, about $869 a month
- Worth it above
- 5B tokens
- a month, against the cheapest API at $0.18 per M
| Quantization | Memory | Smallest cloud GPU | Quality |
|---|---|---|---|
| INT4 (Q4) | 70 GB | A100 · $1.19/h | smallest footprint, minor quality loss |
| INT8 (Q8) | 140 GB | 2× H100 · $4.30/h | near-lossless |
| FP16 (full) | 280 GB | 4× H100 · $8.60/h | reference quality |
At home: in INT4 it fits a Mac Studio M5 Ultra (96 GB unified), about $5,499 once, or $0.2092/h over 3 years, electricity aside; it pays for itself after about 31B tokens against the cheapest API.
Memory ≈ parameters × bytes per parameter × 1.2 · cloud price: on-demand rate, running 24/7 · builds: GGUF, AWQ, MLX, GPTQ, Ollama · break-even calculator
| Provider | Input | Output | Cache ↑ | tok/s | First token | Uptime 24h | Checked |
|---|---|---|---|---|---|---|---|
| DigitalOcean125K ctx · 4K out | $0.06 | $0.42 | $0.012 | 49 | 0.83 s | 100% | Oct 23 |
| Veniceint4 · 125K ctx · 32K out | $0.03 | $0.15 | $0.03 | 42 | 1.45 s | 99.8% | Oct 23 |
| CoreWeavefp4 · 128K ctx | $0.03 | $0.17 | $0.03 | 45 | 0.31 s | 100% | Oct 23 |
| DekaLLMbf16 (mxfp4) · 128K ctx | $0.03 | $0.18 | $0.03 | 89 | 0.44 s | 100% | Oct 23 |
| AkashMLbf16 (mxfp4) · 128K ctx | $0.03 | $0.19 | $0.03 | 134 | 0.38 s | 100% | Oct 23 |
| Crusoebf16 (mxfp4) · 128K ctx | $0.05 | $0.25 | $0.05 | 352 | 0.20 s | 99.9% | Oct 23 |
| Parasailfp4 · 128K ctx | $0.10 | $0.75 | $0.055 | 184 | 0.36 s | 100% | Oct 23 |
| Groqprecision not stated · 128K ctx · 64K out | $0.15 | $0.60 | $0.075 | 343 | 0.16 s | 100% | Oct 23 |
| SiliconFlowfp8 (mxfp4) · 128K ctx · 8K out | $0.15 | $0.60 | $0.075 | 55 | 1.27 s | 95.6% | Oct 23 |
| BaseTenfp4 · 128K ctx | $0.10 | $0.50 | $0.10 | 146 | 0.17 s | 99.7% | Oct 23 |
| Cerebrasfp16/8 · 128K ctx · 40K out | $0.35 | $0.75 | $0.35 | 932 | 0.15 s | 99.7% | Oct 23 |
| DeepInfrabf16 (mxfp4) · 128K ctx | $0.037 | $0.17 | — | 50 | 0.36 s | 100% | Oct 23 |
| Novita AIfp4 · 128K ctx · 32K out | $0.05 | $0.25 | — | — | — | — | Oct 23 |
| Googleprecision not stated · 128K ctx | $0.09 | $0.36 | — | 51 | 0.29 s | 29.7% | Oct 23 |
| Amazon Bedrockprecision not read: host terms · 128K ctx | $0.15 | $0.60 | — | 160 | 0.43 s | 100% | Oct 23 |
| Nebiusfp4 · 128K ctx | $0.15 | $0.60 | — | 305 | 0.29 s | 100% | Oct 23 |
| Phalaprecision not read: host terms · 128K ctx · 32K out | $0.15 | $0.60 | — | 202 | 0.84 s | 99.5% | Oct 23 |
| Together AImxfp4 · 128K ctx | $0.15 | $0.60 | — | 88 | 0.36 s | 99.9% | Oct 23 |
| SambaNovaprecision not stated · 128K ctx | $0.14 | $0.95 | — | 388 | 1.21 s | 99.8% | Oct 23 |
10 of 19 providers · prices per million tokens, read Oct 23, 06:00 UTC · speed: OpenRouter, median of the last 30 min, read Oct 23, 06:00 UTC · uptime: OpenRouter, last 24 h · under each host: weights precision, context and max output as served, the maker's value in brackets when the host's differs
| Provider | Weights | Context | Max output | Tools | Structured outputs | Reasoning |
|---|---|---|---|---|---|---|
| Venice | int4 | 125K | 32K | yes | no | yes |
| CoreWeave | fp4 | 128K | not checked yet | yes | yes | yes |
| DeepInfra | bf16 | 128K | not stated by the host | yes | yes | yes |
| DekaLLM | bf16 | 128K | not checked yet | yes | yes | yes |
| AkashML | bf16 | 128K | not stated by the host | yes | yes | yes |
| Novita AI | fp4 | 128K | 32K | yes | no | yes |
| Crusoe | bf16 | 128K | not stated by the host | yes | yes | yes |
| not stated by the host | 128K | not stated by the host | no | yes | yes | |
| DigitalOcean | not checked yet | 125K | 4K | no | no | yes |
| BaseTen | fp4 | 128K | not checked yet | yes | yes | yes |
| Groq | not stated by the host | 128K | 64K | yes | yes | yes |
| SiliconFlow | fp8 | 128K | 8K | no | yes | yes |
| Amazon Bedrock | not read: host terms | 128K | not read: host terms | no | no | yes |
| Nebius | fp4 | 128K | not stated by the host | yes | yes | yes |
| Phala | not read: host terms | 128K | 32K | yes | yes | yes |
| Together AI | mxfp4 | 128K | not stated by the host | yes | yes | yes |
| Parasail | fp4 | 128K | not read: host terms | yes | yes | yes |
| SambaNova | not stated by the host | 128K | not stated by the host | yes | no | yes |
| Cerebras | fp16/8 | 128K | 40K | yes | yes | yes |
Weights, context and max output: read at each host's own catalogue or docs where its terms allow, otherwise as the host declares to OpenRouter; hover a cell for its source and date · tools, structured outputs, reasoning: as declared to OpenRouter · not verified by us
gpt-oss-120b spec sheet, field by field, with sources
gpt-oss-120b spec sheetmaker-declared
What OpenAI declares about the model itself — read from its own weights repository and documentation, not from any host. What each endpoint actually serves is in the pricing above: the two can differ, and both are true.
- Context window
- 131,072 tokens
- Parameters
- 116,829,156,672≈ 116.8B
- Active experts
- 4 of 128 active per token
- Weight precision
- mxfp4
- License
- Apache 2.0 — open weights, permissivecuratedScope: the licence on the published weights, and nothing else. It is not the maker's acceptable-use policy, which is a separate document, and it is not the contract you sign with whichever host you call — hosts set their own terms, which this index does not read.
- Knowledge cutoff
- June 2024maker's docs
- Modalities
- text
- Max output (official)
- to verify
- Release date
- to verify≈ August 5, 2025 third-party catalogue — not the maker's figure
- Maker lifecycle status
- to verify
Also declared in the weights
- Architecture class
- GptOssForCausalLM
- Layers
- 36
- Vocabulary
- 201,088 tokens
- Weights first published
- August 4, 2025
- Weights last updated
- August 26, 2025
- Paper
- arXiv:2508.10925
openai/gpt-oss-120b · config.json @ b5c939d · read 2026-10-08 · 7 of 10 fields published by the maker
| Date | Provider | Input | Output | Change |
|---|---|---|---|---|
| Oct 9 | AkashML | $0.037 → $0.03 | $0.19 → $0.19 | −3% |
| Sep 30 | AkashML | $0.033 → $0.037 | $0.19 → $0.19 | +2% |
| Sep 29 | AkashML | $0.03 → $0.033 | $0.17 → $0.19 | +10% |
| Sep 15 | DigitalOcean | $0.055 → $0.06 | $0.39 → $0.42 | +9% |
| Sep 10 | SiliconFlow | $0.05 → $0.15 | $0.45 → $0.60 | +50% |
| Sep 5 | DeepInfra | $0.15 → $0.037 | $0.60 → $0.17 | −72% |
Every published change of input or output price, newest first · change on input + output · full series: pricing-history.csv.
| Provider | Chatbot 1,000 in / 500 out | RAG / long context 8,000 in / 500 out | Batch summarize 4,000 in / 1,000 out |
|---|---|---|---|
| Venice | $0.11 | $0.32 | $0.27 |
| CoreWeave | $0.12 | $0.33 | $0.29 |
| DeepInfra | $0.12 | $0.38 | $0.32 |
| DekaLLM | $0.12 | $0.33 | $0.30 |
| AkashML | $0.12 | $0.33 | $0.31 |
| Novita AI | $0.17 | $0.53 | $0.45 |
| Crusoe | $0.17 | $0.53 | $0.45 |
| $0.27 | $0.90 | $0.72 | |
| DigitalOcean | $0.27 | $0.69 | $0.66 |
| BaseTen | $0.35 | $1.05 | $0.90 |
| Groq | $0.45 | $1.50 | $1.20 |
| SiliconFlow | $0.45 | $1.50 | $1.20 |
| Amazon Bedrock | $0.45 | $1.50 | $1.20 |
| Nebius | $0.45 | $1.50 | $1.20 |
| Phala | $0.45 | $1.50 | $1.20 |
| Together AI | $0.45 | $1.50 | $1.20 |
| Parasail | $0.47 | $1.18 | $1.15 |
| SambaNova | $0.61 | $1.59 | $1.51 |
| Cerebras | $0.72 | $3.17 | $2.15 |
What each host bills for 1,000 requests of a representative job
Your own workload
Other OpenAI models · all its models, hosts and history ›
- gpt-oss-20b13 providers · from $0.12
- GPT-5.6 Luna3 providers · from $1.40
- GPT-5.6 Sol3 providers · from $24.00
- GPT-5.6 Terra3 providers · from $14.00
- GPT-4.11 provider · from $10.00
- GPT-4.1 mini1 provider · from $2.00