Every endpoint, named.
| # | Endpoint | Declared | Exact match | Self-agree | Nights | Drift | $ in / out per M | Quality-adjusted | State |
|---|---|---|---|---|---|---|---|---|---|
| 1 | primeintellect/fp4 | fp4 | 96.0% | 95.2% | 9 | stable | $0.042 / $0.090 | $0.056 | Matches reference |
| 2 | near-ai/fp4 | fp4 | 96.1% | 94.4% | 14 | stable | $0.030 / $0.15 | $0.062 | Matches reference |
| 3 | coreweave/fp4 | fp4 | 94.9% | 93.0% | 14 | stable | $0.030 / $0.17 | $0.068 | Matches reference |
| 4 | dekallm/bf16 | bf16 | 96.4% | 94.4% | 14 | stable | $0.030 / $0.18 | $0.070 | Matches reference |
| 5 | deepinfra/bf16 | bf16 | 96.8% | 96.2% | 14 | stable | $0.037 / $0.17 | $0.073 | Matches reference |
| 6 | crusoe/bf16 | bf16 | 97.8% | 96.0% | 14 | stable | $0.050 / $0.25 | $0.10 | Matches reference |
| 7 | mancer/fp8 | fp8 | 95.7% | 93.8% | 14 | stable | $0.055 / $0.25 | $0.11 | Matches reference |
| 8 | sail-research/fp4 | fp4 | 96.1% | 94.5% | 14 | stable | $0.060 / $0.40 | $0.15 | Matches reference |
| 9 | digitalocean | unknown | 92.1% | 89.2% | 14 | stable | $0.060 / $0.42 | $0.16 | Matches reference |
| 10 | google-vertex/global | unknown | 94.2% | 93.0% | 14 | stable | $0.090 / $0.36 | $0.17 | Matches reference |
| 11 | baseten/fp4 | fp4 | 98.0% | 97.8% | 14 | stable | $0.10 / $0.50 | $0.20 | Matches reference |
| 12 | baseten/fp4/us | fp4 | 97.4% | 97.2% | 14 | stable | $0.10 / $0.50 | $0.21 | Matches reference |
| 13 | together | unknown | 97.2% | 97.4% | 14 | stable | $0.15 / $0.60 | $0.27 | Matches reference |
| 14 | groq | unknown | 96.8% | 98.5% | 14 | stable | $0.15 / $0.60 | $0.27 | Matches reference |
| 15 | amazon-bedrock/eu-west-1 | unknown | 96.3% | 94.6% | 14 | stable | $0.15 / $0.60 | $0.27 | Matches reference |
| 16 | parasail/fp4 | fp4 | 96.3% | 94.0% | 14 | stable | $0.10 / $0.75 | $0.27 | Matches reference |
| 17 | amazon-bedrock | unknown | 96.1% | 94.0% | 14 | stable | $0.15 / $0.60 | $0.27 | Matches reference |
| 18 | nebius/fp4 | fp4 | 95.8% | 94.4% | 14 | stable | $0.15 / $0.60 | $0.27 | Matches reference |
| 19 | siliconflow/fp8 | fp8 | 95.4% | 95.3% | 14 | stable | $0.15 / $0.60 | $0.28 | Matches reference |
| 20 | sambanova | unknown | 97.1% | 95.8% | 14 | stable | $0.14 / $0.95 | $0.35 | Matches reference |
| 21 | deepinfra/fp8 | fp8 | 97.4% | 95.7% | 11 | stable | $0.20 / $0.95 | $0.40 | Matches reference |
| 22 | cerebras/fp16 | fp16 | 97.9% | 98.1% | 14 | stable | $0.35 / $0.75 | $0.46 | Matches reference |
| 23 | mara | unknown | 97.1% | 96.7% | 9 | stable | n/a / n/a | n/a | Matches reference, not listed since Nov 4 |
| 24 | akashml/bf16 | bf16 | 95.8% | 93.7% | 8 | stable | n/a / n/a | n/a | Matches reference, not listed since Nov 3 |
| 25 | phala | unknown | 93.5% | 92.4% | 13 | stable | n/a / n/a | n/a | Matches reference |
| 26 | novita/fp4 | fp4 | 92.6% | 90.6% | 7 | stable | n/a / n/a | n/a | Matches reference, not listed since Nov 2 |
| First look: fewer than 7 nights, not ranked | |||||||||
| akashml/fp4 | fp4 | 94.8% | 93.3% | 6 | stable | $0.030 / $0.19 | $0.073 | First look, 6 nights | |
| parasail/fast | fp4 | 96.4% | 91.2% | 5 | stable | $0.15 / $0.75 | $0.31 | First look, 5 nights | |
| Outside the reference | |||||||||
| venice/int4 | int4 | 85.8% | 80.2% | 14 | worsening | $0.030 / $0.15 | $0.070 | Outside reference | |
Cite: Bitdepth · bitdepth.dev/m/gpt-oss-120b · as of Nov 11, 2026
Reference: plurality among endpoints declaring bf16 or fp16. State rule: outside when self-agreement exceeds agreement with the reference by more than 15 points, or when the endpoint's mean on the model's first primary column is more than 8 points below the median of endpoints with 7+ nights; no_data under 7 nights; stale when the data is more than 3 days old.
Cheap and wrong sits low on the left.
Each dot is one endpoint. Across: blended price per million tokens. Up: agreement with the reference.
Cite: Bitdepth · bitdepth.dev/m/gpt-oss-120b#price · as of Nov 11, 2026
Does an endpoint change?
Exact match with the reference each night, one line per endpoint. × marks a night with no row.
Cite: Bitdepth · bitdepth.dev/m/gpt-oss-120b#nightly · as of Nov 11, 2026
Graded by Bitdepth.
An endpoint that matches the reference can show this badge on its docs or pricing page. It links back to its row here, and it updates every night.
Badge code for a matching endpoint, after the first ranked read:
<a href="https://bitdepth.dev/m/gpt-oss-120b"><img src="https://bitdepth.dev/badge/gpt-oss-120b/PROVIDER.svg" alt="Graded by Bitdepth"></a>How this page is measured.
The reference
For each model, the answer most endpoints declaring full precision (bf16 or fp16) give at temperature zero. Where fewer than two declare it, the plurality answer across all endpoints. Method.
Agreement and self-agreement
Agreement is how often an endpoint gives the reference answer. Self-agreement is how often it gives its own answer twice. An endpoint that agrees with itself far more than with the reference is serving something different.
How rank works
Endpoints that match the reference are ranked by quality-adjusted price: the blended price, (3 × input + output) ÷ 4 per million tokens, divided by agreement. A cheap endpoint that misses the reference is never shown as the cheapest.
First look
An endpoint needs 7 nights before it is ranked. Until then it shows its night count and its numbers, with no rank.
Drift
The mean of the last three nights against the first three. A move of more than 10 points reads as improving or worsening; anything smaller is stable.
How to cite
Every chart and table carries a cite line with the date. Copy it with the figure. Wrong model name or missing endpoint? Tell us.