Compare / head-to-head
GLM-5.3vsKimi K2.6
GLM-5.3 leads 4 of 5 shared benchmarks. GLM-5.3 has the larger context window (1M tokens).
Benchmarks from cited public sources; pricing from official pages; status from official provider feeds.
Shared benchmarks
4 – 1
GLM-5.3 leads
Cheaper per token
—
list price, input + output
Larger context
GLM-5.3
1M tokens
Provider uptime (30d)
— · 100%
Z.ai · Moonshot AI
GLM-5.3
Z.ai · released 2026-08-14
text
Kimi K2.6
Moonshot AI · released 2026-04-20
textvision
- Context
- 256K
- Max out
- —
- Input /1M
- —
- Output /1M
- —
- Cached /1M
- —
- Scores
- 14 · 8 core
Quality
Benchmark matrix
5 shared · 19 only GLM-5.3 · 9 only Kimi K2.6| Benchmark | GLM-5.3 | Kimi K2.6 | Δ | Edge |
|---|---|---|---|---|
| Reported by both · 5 | ||||
| GPQA Diamond | 90.9% ↗epoch | 90.8% ↗epoch | +0.1 pt | GLM-5.3 |
| FrontierMath Tier 4 v2 (Epoch AI run) | 29.3% ↗epoch | 25.6% ↗epoch | +3.7 pt | GLM-5.3 |
| FrontierMath Tiers 1-3 v2 (Epoch AI run) | 68.8% ↗epoch | 57.2% ↗epoch | +11.6 pt | GLM-5.3 |
| OTIS Mock AIME 2024-2025 (Epoch AI run) | 91.1% ↗epoch | 96.1% ↗epoch | -5 pt | Kimi K2.6 |
| SimpleQA Verified | 41% ↗epoch | 34.9% ↗epoch | +6.1 pt | GLM-5.3 |
| Only GLM-5.3 reports · 19 | ||||
| Agents' Last Exam | 28.5% ↗ | not reported | — | — |
| Agents' Last Exam (ALE-CLI) | 28.5% ↗ | not reported | — | — |
| AutomationBench v1.0.6 | 48.2% ↗ | not reported | — | — |
| CyberGym | 84.5% ↗ | not reported | — | — |
| DeepSWE v1.1 | 66.9% ↗ | not reported | — | — |
| ExploitBench | 54.4% ↗ | not reported | — | — |
| ExploitGym 2h | 105 ↗ | not reported | — | — |
| ExploitGym 6h | 130 ↗ | not reported | — | — |
| FrontierSWE | 78.1% ↗ | not reported | — | — |
| GDPval-AA v2 | 1769 ↗ | not reported | — | — |
| HLE (with tools) | 62.5% ↗ | not reported | — | — |
| NL2Repo | 58% ↗ | not reported | — | — |
| PostTrainBench | 39.8% ↗ | not reported | — | — |
| ProgramBench (Almost Solved) | 19% ↗ | not reported | — | — |
| SWE-Marathon v1.1 | 42.5% ↗ | not reported | — | — |
| Terminal-Bench 2.1 | 88.2% ↗ | not reported | — | — |
| Terminal-Bench 3.0 | 28.3% ↗ | not reported | — | — |
| Toolathlon Verified | 73% ↗ | not reported | — | — |
| Z.ai Code Bench (max effort) | 34.5% ↗ | not reported | — | — |
| Only Kimi K2.6 reports · 9 | ||||
| AIME 2026 | not reported | 95.8% ↗matharena ⚠ | — | — |
| LMArena Elo | not reported | 1454.9 ↗ | — | — |
| Kimi Claw 24/7 Bench | not reported | 42.9% ↗ | — | — |
| Kimi Code Bench v2 | not reported | 50.9% ↗ | — | — |
| MCP-Atlas | not reported | 69.4% ↗ | — | — |
| MCPMark-Verified | not reported | 72.8% ↗ | — | — |
| MLS-Bench-Lite | not reported | 26.7% ↗ | — | — |
| ProgramBench | not reported | 48.3% ↗ | — | — |
| SWE-bench Verified (Epoch AI run) | not reported | 76.7% ↗epoch | — | — |
Scores tagged "epoch" or "matharena" are independent runs, used only where the lab has not published its own; ⚠ marks rows MathArena flags as released after the competition. Higher is better on every row. Δ is GLM-5.3 minus Kimi K2.6 in the benchmark's own unit. "Not reported" means the lab has not published that figure; it is not a zero. ↗ opens the source.
Specs & pricing
Side by side
Official pricing pages and model cards| Spec | GLM-5.3 | Kimi K2.6 | Edge |
|---|---|---|---|
| Context window | 1M tokens | 256K tokens | GLM-5.3 |
| Max output | 131K tokens | — | — |
| Input price / 1M | $1.4 ↗ | — | — |
| Output price / 1M | $4.4 ↗ | — | — |
| Cached input / 1M | $0.26 ↗ | — | — |
| Input + output / 1M Lower is cheaper. List prices; batch, tool and regional fees excluded. | $5.8 | — | — |
| Modalities | text | text · vision | Kimi K2.6 |
| Released | 2026-08-14 | 2026-04-20 | — |
| Cited benchmark scores | 33 | 14 | — |
Reliability
Provider status
Last 60 days · refreshed daily on this page · live on /statusMore matchups
GLM-5.3 vs …
Models sharing the most benchmarksMore matchups
Kimi K2.6 vs …
Models sharing the most benchmarksBuilt by Respan
Which one wins on your data?
Public benchmarks are a starting point. Run GLM-5.3 and Kimi K2.6 on your own prompts with Respan evals, or route to either through one gateway key with automatic failover.