#79 Mistral Small 4
medium- Cost
- $0.006
- Time
- 47.9s
- Tokens
- 9,857 tok
Mistral
Release: 2026-03-16
Tested on: 2026-04-11 01:44
mistralai/mistral-small-2603::medium
119B total (6B active)MoEOpen source
(medium)
(none)
Summary
Mistral Small 4 scores 5.7 on AI BENCHY and ranks #79. It has N/A reliability, a 50.0% pass rate, $0.034 total cost, and 5.64s average response time.
What makes Mistral Small 4 unique: Its total benchmark cost is unusually low for its score range. It is notably fast compared with similar models.
Model facts
Researched on 2026-08-12
- Parameters
- 119B total (6B active)
- Architecture
- MoE
- Availability
- Open source
- License
- Apache-2.0
The vendor also reports 8B active when embeddings are included; UI uses routed active parameters.
5.7
Consistency
6.8
N/A
$0.034
Total Output Tokens
54,492
Total Input Tokens
0
Input Price
$0.150 / 1M
Output Price
$0.600 / 1M
Flaky tests
7
Flaky tests had mixed outcomes across runs (at least one pass and one fail).
Hamster playing table tennis
Prompt: Create a detailed SVG illustration of a hamster playing table tennis.
Run history
| Tested on | Score | Reliability | Tests Correct | Total Cost | Compare |
|---|---|---|---|---|---|
| 2026-08-14 02:05 Re-test | 5.1 | 10.0 | $0.097 | Compare | |
| 2026-07-16 22:23 New test added | 5.1 | 10.0 | $0.096 | Compare | |
| 2026-06-04 13:43 New test added | 5.3 | 10.0 | $0.068 | Compare | |
| 2026-05-22 00:16 Suite changed | 5.4 | 10.0 | $0.056 | Compare | |
| 2026-04-11 01:44 First recorded run | 5.7 | N/A | $0.034 | Current run |
Run comparison
| Run | Benchmark coverage | Score | Consistency | Reliability | Tests Correct | Flaky tests | Total Output Tokens | Total Input Tokens | Total Cost | Response Time (avg) |
|---|---|---|---|---|---|---|---|---|---|---|
| 2026-04-11 01:44 · First recorded run | 54/54 attempts | 5.7 | 6.8 | N/A | 5/18 | 7 | 54,492 | 0 | $0.034 | 5.64s |
| 2026-08-14 02:05 · Re-test | 66/66 attempts | 5.1 | 7.0 | 10.0 | 5/22 | 8 | 133,597 | 140,559 | $0.097 | 10.80s |
| Difference | — | +0.5 | -0.2 | 0 | -1 | -79105 | -140559 | -$0.063 | -5160ms |
Benchmark coverage differs: 54/54 attempts (Target: 3 repeats per test) versus 66/66 attempts (Target: 3 repeats per test). Totals and repeat-sensitive metrics are not directly comparable.
These two runs used different benchmark suites, so the deltas reflect both model changes and suite changes.
Charts
Choose the first model, then click a second model to open a side-by-side page.
Top Models by Score
Score vs Total Cost
Response Time (avg)
Score vs Response Time (avg)
Total Output Tokens
Score vs Total Output Tokens
Quick Compare
Mistral Small 4mediumvsMistral Small 4noneMistral Small 4mediumvsGranite 4.2 8BlowMistral Small 4mediumvsNemotron 3.5 LightningmediumFree AvailableMistral Small 4mediumvsQwen3 Coder NextnoneMistral Small 4mediumvsGranite 4.2 8BmediumMistral Small 4mediumvsMercury 2.5lowMistral Small 4mediumvsInklingnoneMistral Small 4mediumvsMiMo-V2.5noneMistral Small 4mediumvsDots 3 Note PreviewnoneFree AvailableMistral Small 4mediumvsQwen3.5-9Bnone
Category Breakdown
| Category | Score | Consistency | Tests Correct |
|---|---|---|---|
| Anti-AI Tricks | 5.6 | 3.8 | |
| Coding | 6.7 | 3.5 | |
| Combined | 3.0 | 10.0 | |
| Data parsing and extraction | 7.3 | 5.9 | |
| Domain specific | 5.3 | 7.2 | |
| General Intelligence | 4.8 | 10.0 | |
| Instructions following | 7.3 | 5.8 | |
| Puzzle Solving | 3.4 | 9.7 | |
| Tool Calling | 10.0 | 10.0 |