DeepSeek-R1 vs Kimi K2 (Non-Reasoning)
Wins 2 of 6 areas
Facts · Long documents
Wins 2 of 6 areas
Agents · Following instructions
The two are evenly matched.DeepSeek-R1 is better at facts and long documents; Kimi K2 (Non-Reasoning) at agents and following instructions.
Scores updated · 18 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.
- FactsGetting facts right instead of making them up21DeepSeek-R12 of 3 tests
- Long documentsFinding answers in very long texts10DeepSeek-R11 of 1 test
- AgentsCarrying out multi-step tasks on its own01Kimi K2 (Non-Reasoning)1 of 1 test
- Following instructionsDoing exactly what it is asked01Kimi K2 (Non-Reasoning)1 of 1 test
- CodingWriting and fixing software11Even1 each
- ReasoningHard problems that need careful thinking11Even1 each · 1 tie
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.
Kimi K2 (Non-Reasoning) costs 52% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where DeepSeek-R1 pulls ahead
- Reasons across sets of long documentsAA-LCR+4.7points ahead
- Code for real scientific research problemsSciCode+3.8points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+3.1points ahead
Where Kimi K2 (Non-Reasoning) pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+13.7points ahead
- Hard command-line tasks in a real terminalTerminal-Bench Hard+9.8points ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+5.8points ahead
Every test, side by side
All 18 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingEven
- Terminal-Bench HardKimi K2 (Non-Reasoning) by 9.86.115.9+9.8
- SciCodeDeepSeek-R1 by 3.838.334.5+3.8
ReasoningEven
- GPQA DiamondKimi K2 (Non-Reasoning) by 5.870.876.6+5.8
- Humanity's Last ExamDeepSeek-R1 by 1.18.57.4+1.1
- CritPttie0.60tie
FactsDeepSeek-R1
- AA-Omniscience · Non-hallucinationKimi K2 (Non-Reasoning) by 13.79.723.4+13.7
- Vectara HHEM hallucination ratelower is betterDeepSeek-R1 by 6.611.317.9+6.6
- AA-Omniscience · AccuracyDeepSeek-R1 by 3.130.527.4+3.1
Following instructionsKimi K2 (Non-Reasoning)
- IFBenchKimi K2 (Non-Reasoning) by 2.53941.5+2.5
Other results7 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)Kimi K2 (Non-Reasoning) by 49.711.461.1+49.7
- vectara_avg_summary_lengthDeepSeek-R1 by 34.393.559.2+34.3
- Artificial Analysis Coding IndexKimi K2 (Non-Reasoning) by 10.224.634.8+10.2
- vectara_factual_consistencyDeepSeek-R1 by 6.688.782.1+6.6
- AA-OmniscienceKimi K2 (Non-Reasoning) by 3.9-32.2-28.3+3.9
- vectara_answer_rateKimi K2 (Non-Reasoning) by 1.69798.6+1.6
- AA IntelligenceKimi K2 (Non-Reasoning) by 1.311.412.7+1.3
Questions people ask
Which is better, DeepSeek-R1 or Kimi K2 (Non-Reasoning)?
DeepSeek-R1 and Kimi K2 (Non-Reasoning) each win two of the six areas where both have results. DeepSeek-R1 wins facts and long documents; Kimi K2 (Non-Reasoning) wins agents and following instructions. Kimi K2 (Non-Reasoning) costs 52% less. They are level on coding and reasoning.
Which is better for coding?
Neither. They win 1 coding test each of the 2 both models report.
Which is cheaper?
DeepSeek-R1 costs $2.00 per million input tokens and $4.00 per million output tokens; Kimi K2 (Non-Reasoning) costs $0.57 and $2.30. That makes Kimi K2 (Non-Reasoning) about 52% cheaper for the same work.
How do you compare the two?
We use the 18 benchmark tests both models have published scores on. The verdict counts the 11 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 7 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.