Serving provider
Neurometric
Neurometric-managed model deployments, including task-optimized models. Published task evidence shows where each model has been evaluated.
← All providers
Serving capacity can use Amazon Bedrock, GCP, Together AI, and self-hosted GPUs. Availability varies by model.
59 models with current observed prices · 7 with published task results
| Model | Evidence | Cheapest passing | Median quality | Input / output per 1M |
|---|---|---|---|---|
| GPT-5.4 Frontier | 263 tasks | 69 | 93.0% | $2.5 / $15 |
| DeepSeek V3 Mid | 259 tasks | 12 | 66.7% | $0.58 / $1.68 |
| Qwen3 235B A22B Mid · 235B (22B active) | 259 tasks | 10 | 54.0% | $0.22 / $0.88 |
| Claude Opus 4.7 Frontier | 259 tasks | 5 | 79.0% | $5.5 / $27.5 |
| Mistral Large 2407 Mid · 123B | 259 tasks | 5 | 66.0% | $3 / $9 |
| Gemini 3.1 Pro Preview Frontier | 259 tasks | 0 | 54.0% | $2 / $12 |
| Nemotron Nano 9B v2 Small · 9B | 259 tasks | 0 | 31.0% | $0.06 / $0.23 |
| azure-gpt-4o Unclassified | Not yet measured | — | — | $2.5 / $10 |
| azure-gpt-4o-mini Unclassified | Not yet measured | — | — | $0.15 / $0.6 |
| azure-gpt-5-nano Unclassified | Not yet measured | — | — | $0.05 / $0.4 |
| claude-haiku-4-5 Unclassified | Not yet measured | — | — | $1 / $5 |
| claude-opus-4-5 Unclassified | Not yet measured | — | — | $5 / $25 |
| claude-opus-4-6 Unclassified | Not yet measured | — | — | $5.5 / $27.5 |
| claude-sonnet-3-7 Unclassified | Not yet measured | — | — | $3 / $15 |
| claude-sonnet-4-0 Unclassified | Not yet measured | — | — | $3 / $15 |
| claude-sonnet-4-5 Unclassified | Not yet measured | — | — | $3 / $15 |
| claude-sonnet-4-6 Unclassified | Not yet measured | — | — | $3.3 / $16.5 |
| deepseek-r1-v1 Unclassified | Not yet measured | — | — | $1.35 / $5.4 |
| gemini-2.5-flash Unclassified | Not yet measured | — | — | $0.3 / $2.5 |
| gemini-2.5-flash-lite Unclassified | Not yet measured | — | — | $0.1 / $0.4 |