NVIDIAPhi-4-Mini
Best configuration ranks #593 of 610 on the BenchLeader Index at 35.5 ±8.7. Last measured 27 Feb 2025stale: no new result in six months. Released 27 Feb 2025.
- Blended price
- $0.131/M
- $0.075 in · $0.300 out
- Output speed
- 45 tok/s
- First answer
- 0.84 s
- first token 0.84 s
- Context
- 131k
- Overall index36
- Coding5
- Agents & tools34
- Knowledge35
- Instruction following26
- Long context33
- Composite37
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. “Default” means the publisher did not say which setting was used.
| Effort | Index | Rank | Speed | First answer | Chat reply cost | Categories |
|---|---|---|---|---|---|---|
| thinking | – | – | 45 tok/s | 0.84 s | $0.0001 | |
| defaultbest | 35.5 | #593 | 45 tok/s | 0.84 s | $0.0001 | Agents & tools 34 · Coding 5 · Composite 37 · Instruction following 26 · Knowledge 35 · Long context 33 |
Benchmark results
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 | default | Source |
|---|---|---|
| GPQA Diamond (AA)not in index | 33.1%#476 | Artificial Analysis |
| Humanity's Last Exam (AA)not in index | 4.5%#423 | Artificial Analysis |
Agents & tools
| Benchmark | default | Source |
|---|---|---|
| Terminal-Bench Hard | 0.0%#345 | Artificial Analysis |
| τ²-Bench Telecom (AA)not in index | 8.2%#371 | Artificial Analysis |
Knowledge
| Benchmark | default | Source |
|---|---|---|
| AA-Omniscience | -61.1#395 | Artificial Analysis |
Instruction following
| Benchmark | default | Source |
|---|---|---|
| IFBench | 21.1%#389 | Artificial Analysis |
Long context
| Benchmark | default | Source |
|---|---|---|
| AA-LCR | 15.3%#380 | Artificial Analysis |
Composite
| Benchmark | default | Source |
|---|---|---|
| AA Intelligence Index | 6.3#450 | Artificial Analysis |
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. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0001 | – | 7.5 s |
| Summarise a 30-page report | 12,000 / 600 | $0.0011 | – | 14.1 s |
| Code edit | 6,000 / 1,500 | $0.0009 | – | 34.0 s |
| Agentic coding session | 60,000 / 4,000 | $0.0057 | – | 1.5 min |
| Structured extraction | 2,000 / 200 | $0.0002 | – | 5.3 s |
See also
Data as of 9 Sept 2026. Compare these configurations.