GLM 5.3 vs Kimi K3
Different efforts: low · max · Compare all at max
Humanity's Last Exam Humanity's Last Exam: Very hard expert questions across many fields, written to stump today’s models
GLM 5.336.6
Kimi K346.9
Arena Text · overall Arena Text · overall: Human pairwise preferences in the stated text vote pool
GLM 5.3 · max1,479
Kimi K31,487
Arena Text · Chinese Arena Text · chinese: Human pairwise preferences in the stated text vote pool
GLM 5.3 · max1,528.8
Kimi K31,528.2
Arena Text · English Arena Text · english: Human pairwise preferences in the stated text vote pool
GLM 5.3 · max1,480
Kimi K31,481.3
Arena Text · French Arena Text · french: Human pairwise preferences in the stated text vote pool
GLM 5.3 · max1,501.6
Kimi K31,488
Arena Text · German Arena Text · german: Human pairwise preferences in the stated text vote pool
GLM 5.3 · max1,492.9
Kimi K31,489.5
Arena Text · instruction following Arena Text · instruction following: Human pairwise preferences in the stated text vote pool
GLM 5.3 · max1,476.3
Kimi K31,483.4
Arena Text · Japanese Arena Text · japanese: Human pairwise preferences in the stated text vote pool
GLM 5.3 · max1,454.7
Kimi K31,489.3
Arena Text · Korean Arena Text · korean: Human pairwise preferences in the stated text vote pool
GLM 5.3 · max1,466.3
Kimi K31,460
Arena Text · non-English Arena Text · non english: Human pairwise preferences in the stated text vote pool
GLM 5.3 · max1,457.7
Kimi K31,464.3
Arena Text · Polish Arena Text · polish: Human pairwise preferences in the stated text vote pool
GLM 5.3 · max1,468.4
Kimi K31,459.9
Arena Text · Russian Arena Text · russian: Human pairwise preferences in the stated text vote pool
GLM 5.3 · max1,464.1
Kimi K31,481.1
Arena Text · Spanish Arena Text · spanish: Human pairwise preferences in the stated text vote pool
GLM 5.3 · max1,464.4
Kimi K31,470.8
AA Omniscience Index AA Omniscience Index: Scores factual knowledge while penalizing incorrect answers to encourage admitting uncertainty
GLM 5.3-8.4
Kimi K319.7
SimpleQA Verified v1 SimpleQA Verified v1: Accuracy on short factual questions with verified answers, answered without web access
GLM 5.3 · max41
Kimi K350.6
CritPt (physics) CritPt (physics): Research-level physics problems that need many careful reasoning steps
GLM 5.314.6
Kimi K323.4
AA Intelligence Index 4.3.2 AA Intelligence Index 4.3.2: Artificial Analysis combines tests of knowledge, reasoning, coding and tool use into one score
GLM 5.334.3
Kimi K343.6
AA Omniscience · Accuracy AA Omniscience · Accuracy: Measures correct answers on the AA Omniscience factual knowledge test
GLM 5.333.9
Kimi K347.6
AA Omniscience · Hallucination Rate AA Omniscience · Hallucination Rate: Measures how often attempted answers are incorrect on the AA Omniscience factual knowledge test · lower is better
GLM 5.364
Kimi K353.2
AA-LCR (long context) 1.1 AA-LCR (long context) 1.1: Questions that need reading and connecting facts across very long documents (around 100K tokens)
GLM 5.372.7
Kimi K388.7
GDP.pdf · AA all-pass GDP.pdf · AA all-pass: Models reason over professional PDF documents; an attempt passes only when every atomic criterion passes
GLM 5.38.8
Kimi K322
MLCR-AA MLCR-AA: Models synthesize long medical records; the headline credits concise answers judged both accurate and complete
GLM 5.3 · max48.3
Kimi K338.3
Chess Puzzles Chess Puzzles: Models choose the best next move from a novel chess position
GLM 5.3 · max37
Kimi K336
FrontierMath Tier 4 (v2) FrontierMath Tier 4 (v2): Models solve expert-written advanced mathematics problems on the named tier and edition
GLM 5.3 · max29.3
Kimi K339
FrontierMath Tiers 1–3 (v2) FrontierMath Tiers 1–3 (v2): Models solve expert-written advanced mathematics problems on the named tier and edition
GLM 5.3 · max68.8
Kimi K372.2
Mystery Game Puzzles Mystery Game Puzzles: Models choose the best move in puzzle positions from an undisclosed game
GLM 5.3 · max33
Kimi K326
Vals LegalBench Vals LegalBench: Legal reasoning and legal-language understanding on LegalBench
GLM 5.384.8
Kimi K386.2
Vals MedCode Vals MedCode: Medical coding accuracy on patient encounters
GLM 5.342.9
Kimi K349.4
Vals MedScribe Vals MedScribe: Accuracy of clinical notes produced from consultation transcripts
GLM 5.388.8
Kimi K388.1
Vals Legal Research Bench Vals Legal Research Bench: Grounded legal research; a task passes only if all criteria pass
GLM 5.349
Kimi K346.2
ARC-AGI-1 · public evaluation ARC-AGI-1 · public-eval: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit
GLM 5.3—
Kimi K394.9
ARC-AGI-1 · semi-private ARC-AGI-1 · semi-private: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit
GLM 5.3—
Kimi K394.5
ARC-AGI-2 · public evaluation ARC-AGI-2 · public-eval: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit
GLM 5.3—
Kimi K361.7
ARC-AGI-2 · semi-private ARC-AGI-2 · semi-private: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit
GLM 5.3—
Kimi K360.4
FrontierMath Erdos FrontierMath Erdos: Models solve expert-written advanced mathematics problems on the named tier and edition
GLM 5.3—
Kimi K30
MathArena · Apex MathArena · apex · Apex: Answer accuracy on the named mathematics competition under the recorded effort
GLM 5.3—
Kimi K365.6
MathArena · Apex Shortlist MathArena · apex · Apex Shortlist: Answer accuracy on the named mathematics competition under the recorded effort
GLM 5.3—
Kimi K392
MathArena · ArXivMath · April 2026 MathArena · arxiv · 04/2026: Answer accuracy on the named mathematics competition under the recorded effort
GLM 5.3—
Kimi K354.2
MathArena · ArXivMath · August 2026 MathArena · arxiv · 08/2026: Answer accuracy on the named mathematics competition under the recorded effort
GLM 5.3—
Kimi K338.6
MathArena · ArXivMath · June 2026 MathArena · arxiv · 06/2026: Answer accuracy on the named mathematics competition under the recorded effort
GLM 5.3—
Kimi K371.5
MathArena · ArXivMath · May 2026 MathArena · arxiv · 05/2026: Answer accuracy on the named mathematics competition under the recorded effort
GLM 5.3—
Kimi K361.7
MathArena · BrokenArXiv · April 2026 MathArena · arxiv_false · 04/2026: Answer accuracy on the named mathematics competition under the recorded effort
GLM 5.3—
Kimi K359.4
MathArena · BrokenArXiv · August 2026 MathArena · arxiv_false · 08/2026: Answer accuracy on the named mathematics competition under the recorded effort
GLM 5.3—
Kimi K361.9
MathArena · BrokenArXiv · June 2026 MathArena · arxiv_false · 06/2026: Answer accuracy on the named mathematics competition under the recorded effort
GLM 5.3—
Kimi K351.9
MathArena · BrokenArXiv · May 2026 MathArena · arxiv_false · 05/2026: Answer accuracy on the named mathematics competition under the recorded effort
GLM 5.3—
Kimi K342.5
MathArena · Kangaroo 2025 1-2 MathArena · kangaroo · Kangaroo 2025 1-2: Answer accuracy on the named mathematics competition under the recorded effort
GLM 5.3—
Kimi K393.8
MathArena · Kangaroo 2025 11-12 MathArena · kangaroo · Kangaroo 2025 11-12: Answer accuracy on the named mathematics competition under the recorded effort
GLM 5.3—
Kimi K398.3
MathArena · Kangaroo 2025 3-4 MathArena · kangaroo · Kangaroo 2025 3-4: Answer accuracy on the named mathematics competition under the recorded effort
GLM 5.3—
Kimi K377.1
MathArena · Kangaroo 2025 5-6 MathArena · kangaroo · Kangaroo 2025 5-6: Answer accuracy on the named mathematics competition under the recorded effort
GLM 5.3—
Kimi K381.7
MathArena · Kangaroo 2025 7-8 MathArena · kangaroo · Kangaroo 2025 7-8: Answer accuracy on the named mathematics competition under the recorded effort
GLM 5.3—
Kimi K396.7
MathArena · Kangaroo 2025 9-10 MathArena · kangaroo · Kangaroo 2025 9-10: Answer accuracy on the named mathematics competition under the recorded effort
GLM 5.3—
Kimi K3100
EQ-Bench 4 Active emotional and social intelligence in multi-turn conversations with 120 simulated personas
GLM 5.3—
Kimi K31,339.3
| Benchmark | GLM 5.3 | Kimi K3 |
|---|---|---|
| Humanity's Last Exam | 36.6 | 46.9 |
| Arena Text · overall | 1,479 · max | 1,487 |
| Arena Text · Chinese | 1,528.8 · max | 1,528.2 |
| Arena Text · English | 1,480 · max | 1,481.3 |
| Arena Text · French | 1,501.6 · max | 1,488 |
| Arena Text · German | 1,492.9 · max | 1,489.5 |
| Arena Text · instruction following | 1,476.3 · max | 1,483.4 |
| Arena Text · Japanese | 1,454.7 · max | 1,489.3 |
| Arena Text · Korean | 1,466.3 · max | 1,460 |
| Arena Text · non-English | 1,457.7 · max | 1,464.3 |
| Arena Text · Polish | 1,468.4 · max | 1,459.9 |
| Arena Text · Russian | 1,464.1 · max | 1,481.1 |
| Arena Text · Spanish | 1,464.4 · max | 1,470.8 |
| AA Omniscience Index | -8.4 | 19.7 |
| SimpleQA Verified v1 | 41 · max | 50.6 |
| CritPt (physics) | 14.6 | 23.4 |
| AA Intelligence Index 4.3.2 | 34.3 | 43.6 |
| AA Omniscience · Accuracy | 33.9 | 47.6 |
| AA Omniscience · Hallucination Rate | 64 | 53.2 |
| AA-LCR (long context) 1.1 | 72.7 | 88.7 |
| GDP.pdf · AA all-pass | 8.8 | 22 |
| MLCR-AA | 48.3 · max | 38.3 |
| Chess Puzzles | 37 · max | 36 |
| FrontierMath Tier 4 (v2) | 29.3 · max | 39 |
| FrontierMath Tiers 1–3 (v2) | 68.8 · max | 72.2 |
| Mystery Game Puzzles | 33 · max | 26 |
| Vals LegalBench | 84.8 | 86.2 |
| Vals MedCode | 42.9 | 49.4 |
| Vals MedScribe | 88.8 | 88.1 |
| Vals Legal Research Bench | 49 | 46.2 |
| ARC-AGI-1 · public evaluation | — | 94.9 |
| ARC-AGI-1 · semi-private | — | 94.5 |
| ARC-AGI-2 · public evaluation | — | 61.7 |
| ARC-AGI-2 · semi-private | — | 60.4 |
| FrontierMath Erdos | — | 0 |
| MathArena · Apex | — | 65.6 |
| MathArena · Apex Shortlist | — | 92 |
| MathArena · ArXivMath · April 2026 | — | 54.2 |
| MathArena · ArXivMath · August 2026 | — | 38.6 |
| MathArena · ArXivMath · June 2026 | — | 71.5 |
| MathArena · ArXivMath · May 2026 | — | 61.7 |
| MathArena · BrokenArXiv · April 2026 | — | 59.4 |
| MathArena · BrokenArXiv · August 2026 | — | 61.9 |
| MathArena · BrokenArXiv · June 2026 | — | 51.9 |
| MathArena · BrokenArXiv · May 2026 | — | 42.5 |
| MathArena · Kangaroo 2025 1-2 | — | 93.8 |
| MathArena · Kangaroo 2025 11-12 | — | 98.3 |
| MathArena · Kangaroo 2025 3-4 | — | 77.1 |
| MathArena · Kangaroo 2025 5-6 | — | 81.7 |
| MathArena · Kangaroo 2025 7-8 | — | 96.7 |
| MathArena · Kangaroo 2025 9-10 | — | 100 |
| EQ-Bench 4 | — | 1,339.3 |
Specs
| Spec | GLM 5.3 | Kimi K3 |
|---|---|---|
| Input, per 1M | !$1.40 | $3 |
| Output, per 1M | !$4.40 | $15 |
| Cost per task? | $0.85 | $2.00 |
| Speed, tok/s | 126 | 112 |
| Context | 1M | 1M |
| Weights | Open | Open |
| Released | Aug 2026 | Jul 2026 |
| Memory (4-bit) | 452 GB | 1,668 GB |
| Licence | glm-5.3 | kimi-k3 |
| Commercial use | Restricted | Restricted |
| See images | — | ✓ |
| Read PDFs | — | Partial |
| Hear audio | — | — |
| Watch video | — | ✓ |
| Call tools | ✓ | ✓ |
| Force a tool call | — | ✓ |
| Return strict JSON | ✓ | ✓ |
| Search the web | ? | Partial |
| Run code | ? | ? |
| Use a computer | ? | ? |
What breaks when you switch
GLM 5.3 → Kimi K3
Memory (4-bit): 452 GB → 1,668 GB
Sampling parameters are fixed›
temperature, top_p, n and penalty values are fixed; omit them from requests.
No public image URLs›
Vision input rejects public image URLs; use base64 or ms:// file references in a content array.
Web search is being updated›
Web search is not recommended for near-term production workflows; avoid depending on it in production.
Thinking mode is always enabled›
K3 always has thinking enabled; control effort with the top-level reasoning_effort field using low, high or max.
Kimi K3 → GLM 5.3
No image input
Reasoning cannot be disabled›
Disabling reasoning is unsupported; keep reasoning enabled.
tool_choice auto · Z.ai API›
The first-party chat reference exposes only auto for tool_choice.
Complete unmodified reasoning_content›
Preserved thinking defaults on for Coding Plan and off for the standard API; when enabled, return complete unmodified reasoning_content in its original order to protect quality and cache hits.
Only one stop word›
Only one stop word is supported, so do not rely on a list of alternative stop sequences.