Claude 3 Haiku vs Devstral Small
Wins 2 of 6 areas
Facts · Following instructions
Wins 4 of 6 areas
Coding · Agents · Reasoning · Long documents
Devstral Small is the stronger all-rounder.Claude 3 Haiku is better at facts.
Scores updated · 15 tests both models report · How we compare
Where each one wins
Tests won in each of the six areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- CodingWriting and fixing software02Devstral Small2 of 2 tests
- ReasoningHard problems that need careful thinking01Devstral Small1 of 3 tests · 2 ties
- AgentsCarrying out multi-step tasks on its own01Devstral Small1 of 1 test
- Long documentsFinding answers in very long texts01Devstral Small1 of 1 test
- FactsGetting facts right instead of making them up20Claude 3 Haiku2 of 2 tests
- Following instructionsDoing exactly what it is asked10Claude 3 Haiku1 of 1 test
Agents, long documents and following instructions rest on a single test each.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Devstral Small pulls ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+6points ahead
- Code for real scientific research problemsSciCode+5.9points ahead
- Hard command-line tasks in a real terminalTerminal-Bench Hard+5.3points ahead
Where Claude 3 Haiku pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+5.9points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+4.5points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+1.7points ahead
Every test, side by side
All 15 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingDevstral Small
- SciCodeDevstral Small by 5.918.624.5+5.9
- Terminal-Bench HardDevstral Small by 5.30.86.1+5.3
ReasoningDevstral Small
- GPQA DiamondDevstral Small by 637.443.4+6
- Humanity's Last Examtie4.14tie
- CritPttie00tie
FactsClaude 3 Haiku
- AA-Omniscience · Non-hallucinationClaude 3 Haiku by 5.919.513.6+5.9
- AA-Omniscience · AccuracyClaude 3 Haiku by 1.717.615.9+1.7
Following instructionsClaude 3 Haiku
- IFBenchClaude 3 Haiku by 4.536.131.6+4.5
Other results5 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- AA Agentic IndexDevstral Small by 18.1725.1+18.1
- τ²-Bench Telecom (AA run)Devstral Small by 16.921.138+16.9
- AA-OmniscienceClaude 3 Haiku by 8.2-48.6-56.8+8.2
- Artificial Analysis Coding IndexDevstral Small by 5.56.712.2+5.5
- AA IntelligenceDevstral Small by 3.15.68.7+3.1
Questions people ask
Which is better, Claude 3 Haiku or Devstral Small?
Devstral Small wins four of the six areas where both have results: coding, agents, reasoning and long documents. Claude 3 Haiku wins facts and following instructions.
Which is better for coding?
Devstral Small. It wins 2 of the 2 coding tests both models report; Claude 3 Haiku wins none.
How do you compare the two?
We use the 15 benchmark tests both models have published scores on. The verdict counts the 10 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 5 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.