Claude 3 Haiku vs Tri-21B-Think
Wins 2 of 5 areas
Facts · Long documents
Wins 2 of 5 areas
Reasoning · Following instructions
The two are evenly matched.Claude 3 Haiku is better at facts and long documents; Tri-21B-Think at reasoning and following instructions.
Scores updated · 11 tests both models report · How we compare
Where each one wins
Tests won in each of the five 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.
- FactsGetting facts right instead of making them up20Claude 3 Haiku2 of 2 tests
- Long documentsFinding answers in very long texts10Claude 3 Haiku1 of 1 test
- ReasoningHard problems that need careful thinking02Tri-21B-Think2 of 3 tests · 1 tie
- Following instructionsDoing exactly what it is asked01Tri-21B-Think1 of 1 test
- CodingWriting and fixing software00Even0 each · 1 tie
Coding, 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 Claude 3 Haiku pulls ahead
- Reasons across sets of long documentsAA-LCR+12points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+5.3points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+4.9points ahead
Where Tri-21B-Think pulls ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+22.7points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+18.5points ahead
- Very hard expert questions across many subjectsHumanity's Last Exam+1.8points ahead
Every test, side by side
All 11 tests both models report. The winning score is in its model's colour; marks a score checked independently.
ReasoningTri-21B-Think
- GPQA DiamondTri-21B-Think by 22.737.460.1+22.7
- Humanity's Last ExamTri-21B-Think by 1.84.15.9+1.8
- CritPttie00.3tie
FactsClaude 3 Haiku
- AA-Omniscience · AccuracyClaude 3 Haiku by 5.317.612.3+5.3
- AA-Omniscience · Non-hallucinationClaude 3 Haiku by 4.919.514.6+4.9
Following instructionsTri-21B-Think
- IFBenchTri-21B-Think by 18.536.154.6+18.5
Other results3 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.
- τ²-Bench Telecom (AA run)Tri-21B-Think by 59.921.181+59.9
- AA-OmniscienceClaude 3 Haiku by 13.9-48.6-62.5+13.9
- AA IntelligenceTri-21B-Think by 3.45.69+3.4
Questions people ask
Which is better, Claude 3 Haiku or Tri-21B-Think?
Claude 3 Haiku and Tri-21B-Think each win two of the five areas where both have results. Claude 3 Haiku wins facts and long documents; Tri-21B-Think wins reasoning and following instructions. They are level on coding.
Which is better for coding?
Neither. All 1 coding tests both models report are ties.
How do you compare the two?
We use the 11 benchmark tests both models have published scores on. The verdict counts the 8 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 3 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.