Agentic
Like-for-like- GLM-4.7
- 45.7
- Kimi K2.5
- 55.0
- Weighted basis
- 2 vs 2 rows
- Reading
- Kimi K2.5 leads
GLM-4.7 vs Kimi K2.5
Updated September 3, 2026. Public scores include evidence status and uncertainty. They are not guarantees for a specific workload.
Decision reading
GLM-4.7 has the higher public score estimate, 60.91 versus 58.81, but the 90% score intervals overlap. Treat that as a lead, not a settled winner.
11 results are shared. Category rows based on different benchmark sets are marked directional and do not name a winner.
Which one for your work
Recommendations appear only when a shared evidence basis or an explicit operating constraint supports the call. Secondary and unsupported use cases stay disclosed below the initial list.
Agentic work
Tool use, computer use, and multi-step task completion
Kimi K2.5
Kimi K2.5 leads on the same 2 weighted benchmark rows.
Confidence: limited
Long documents
Prompts that approach the documented context limit
Kimi K2.5
Kimi K2.5 has the larger documented context window.
Confidence: documented
Show secondary and unsupported calls
Coding work
Code generation, repair, and software-engineering tasks
Not enough matched evidence
The category averages use different weighted benchmark sets, so they are directional rather than like-for-like.
Confidence: limited
Chat turn cost
1K fresh input + 500 output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Confidence: listed-rates
Cache-heavy agent loop cost
200K cached + 20K fresh input + 10K output tokens
Not enough matched evidence
The page does not recommend a cost winner because at least one model cannot fit the stated workload in one request. GLM-4.7 does not fit this workload in one request. Kimi K2.5 has no published cached-input rate, so cached tokens use its listed input rate. GLM-4.7 has no comparable published API token rate.
Confidence: rate-fallback
Repository review cost
50K fresh input + 3K output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Confidence: listed-rates
What is actually comparable
Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.
2 GLM-4.7 only
11 Shared
34 Kimi K2.5 only
- Shared results
- 11
- GLM-4.7 only
- 2
- Kimi K2.5 only
- 34
- Like-for-like categories
- 1 / 8
3 categories use different evidence sets. Those rows remain visible for coverage context but do not name a winner.
Category results, on a stated basis
Each row states whether both averages use the same weighted benchmark set. Directional and not-comparable rows remain visible, but they never receive a winner in this template.
Coding
Directional only- GLM-4.7
- 75.4
- Kimi K2.5
- 59.4
- Weighted basis
- 3 vs 4 rows
- Reading
- Directional only
Knowledge
Directional only- GLM-4.7
- 51.8
- Kimi K2.5
- 56.9
- Weighted basis
- 3 vs 4 rows
- Reading
- Directional only
Math
Directional only- GLM-4.7
- 1.8
- Kimi K2.5
- 60.6
- Weighted basis
- 2 vs 4 rows
- Reading
- Directional only
Reasoning
Not comparable- GLM-4.7
- Not measured
- Kimi K2.5
- 61.0
- Weighted basis
- 0 vs 1 rows
- Reading
- Not comparable
Multilingual
Not comparable- GLM-4.7
- Not measured
- Kimi K2.5
- 82.3
- Weighted basis
- 0 vs 1 rows
- Reading
- Not comparable
Multimodal
Not comparable- GLM-4.7
- Not measured
- Kimi K2.5
- 78.5
- Weighted basis
- 0 vs 1 rows
- Reading
- Not comparable
Instruction following
Not comparable- GLM-4.7
- Not measured
- Kimi K2.5
- 93.9
- Weighted basis
- 0 vs 1 rows
- Reading
- Not comparable
| Category | GLM-4.7 | Kimi K2.5 | Weighted basis | Reading |
|---|---|---|---|---|
| Agentic | 45.7 | 55.0 | Like-for-like2 vs 2 rows | Kimi K2.5 leads |
| Coding | 75.4 | 59.4 | Directional only3 vs 4 rows | Directional only |
| Knowledge | 51.8 | 56.9 | Directional only3 vs 4 rows | Directional only |
| Math | 1.8 | 60.6 | Directional only2 vs 4 rows | Directional only |
| Reasoning | Not measured | 61.0 | Not comparable0 vs 1 rows | Not comparable |
| Multilingual | Not measured | 82.3 | Not comparable0 vs 1 rows | Not comparable |
| Multimodal | Not measured | 78.5 | Not comparable0 vs 1 rows | Not comparable |
| Instruction following | Not measured | 93.9 | Not comparable0 vs 1 rows | Not comparable |
Shape of the matched evidence
Only shared public evidence is shown. Sparse evidence stays a ruled list rather than being closed into a radar shape.
Too few matched category axes support a radar. The ruled list below shows only shared benchmark results; positions use each benchmark’s normalized display scale when available.
FrontierMath v2 (Tiers 1-3)
Math
Terminal-Bench 2.0
Agentic
BrowseComp
Agentic
HLE
Knowledge
FrontierMath v2 (Tier 4)
Math
What each workload costs
Three fixed token mixes turn per-token rates into comparable decisions. Each scenario states context fit and whether cached input had to fall back to the published list-input rate.
Chat turn
1K fresh input + 500 output tokens
- GLM-4.7
- Self-hosted; infrastructure cost varies
- Fits in one request
- Kimi K2.5
- $0.0021
- Fits in one request
GLM-4.7 has no comparable published API token rate.
Repository review
50K fresh input + 3K output tokens
- GLM-4.7
- Self-hosted; infrastructure cost varies
- Fits in one request
- Kimi K2.5
- $0.039
- Fits in one request
GLM-4.7 has no comparable published API token rate.
Cache-heavy agent loop
200K cached + 20K fresh input + 10K output tokens
- GLM-4.7
- Self-hosted; infrastructure cost varies
- Does not fit in one request
- Cached-input rate unavailable
- Kimi K2.5
- $0.162
- Fits in one request
- Cached input priced at the published list-input rate
GLM-4.7 does not fit this workload in one request. Kimi K2.5 has no published cached-input rate, so cached tokens use its listed input rate. GLM-4.7 has no comparable published API token rate.
Specification differences
Sourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.
SpecificationGLM-4.7Kimi K2.5
Context window
Maximum documented context; output-token limits may be lower.
GLM-4.7
200K
Kimi K2.5
256K
API model ID
GLM-4.7
Not sourced
Kimi K2.5
Not sourced
Cached-input rate
A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.
GLM-4.7
No comparable hosted API rate
Kimi K2.5
Not published
Documented inputs
GLM-4.7
Not sourced
Kimi K2.5
Not sourced
Documented outputs
GLM-4.7
Not sourced
Kimi K2.5
Not sourced
Provider availability
GLM-4.7
Not sourced
Kimi K2.5
Not sourced
Reasoning profile
GLM-4.7
Reasoning
Kimi K2.5
Non-Reasoning
Weight access
GLM-4.7
Open Weight
Kimi K2.5
Open Weight
License
GLM-4.7
Open Weight
Kimi K2.5
Open Weight
Release date
GLM-4.7
2025-10-01
Kimi K2.5
2026-02-01
If you already use one of these models
- Deployment change
- The models list different providers, so authentication, endpoint behavior, limits, and feature support may change.
- Quality signal
- GLM-4.7 has the higher public score estimate, 60.91 versus 58.81, but the 90% score intervals overlap.
- Workload cost
- A complete comparable API-rate estimate is not available for both models.
- Context tradeoff
- Kimi K2.5 has the larger documented window (256K).
Run the same representative tasks against both endpoints before changing production traffic.
Self-host vs API cost
Estimates at 50,000 req/day · 1000 tokens/req average.
GLM-4.7
API / mo$0
Self-host / moNot listed
Break-even—
Proprietary model — self-hosting not applicable.
Kimi K2.5
API / mo$2,700
Self-host / mo$5,221
Break-even132M/day
Benchmark evidence
The full public result ledger is available for audit without forcing a wide desktop table onto a phone.
Browse raw public benchmark evidence47 rows
Agentic
Terminal-Bench 2.0
GLM-4.741%SourceKimi K2.550.8%SourceKimi K2.5 leads this result
BrowseComp
GLM-4.752%SourceKimi K2.560.6%SourceKimi K2.5 leads this result
VITA-Bench
GLM-4.715.5%SourceKimi K2.5—Not directly comparable
Gert Labs
Shared sourceGLM-4.739.95%Kimi K2.545.88%Kimi K2.5 leads this result
Claw-Eval
GLM-4.7—Kimi K2.552.3%SourceNot directly comparable
QwenClawBench
GLM-4.7—Kimi K2.554.3%SourceNot directly comparable
τ³-bench results
GLM-4.7—Kimi K2.565.7%SourceNot directly comparable
DeepSearchQA
GLM-4.7—Kimi K2.577.1%SourceNot directly comparable
DeepPlanning
GLM-4.7—Kimi K2.514.4%SourceNot directly comparable
Toolathlon
GLM-4.7—Kimi K2.527.8%SourceNot directly comparable
MCP Atlas
GLM-4.7—Kimi K2.529.5%SourceNot directly comparable
MCP-Tasks
GLM-4.7—Kimi K2.559.1%SourceNot directly comparable
WideResearch
GLM-4.7—Kimi K2.572.7%SourceNot directly comparable
ResearchClawBench
GLM-4.7—Kimi K2.514.0%SourceNot directly comparable
JobBench
GLM-4.7—Kimi K2.58.7%SourceNot directly comparable
Coding
SWE-bench Verified
GLM-4.773.8%SourceKimi K2.576.8%SourceKimi K2.5 leads this result
LiveCodeBench
GLM-4.784.9%SourceKimi K2.5—Not directly comparable
SWE-Rebench
GLM-4.758.7%SourceKimi K2.558.5%SourceGLM-4.7 leads this result
SWE-bench Verified*
GLM-4.7—Kimi K2.570.8%SourceNot directly comparable
LiveCodeBench v6
GLM-4.7—Kimi K2.585.0%SourceNot directly comparable
SWE-bench Pro
GLM-4.7—Kimi K2.550.7%SourceNot directly comparable
SWE Multilingual
GLM-4.7—Kimi K2.573%SourceNot directly comparable
React Native Evals
GLM-4.7—Kimi K2.577.2%SourceNot directly comparable
SciCode
GLM-4.7—Kimi K2.548.7%SourceNot directly comparable
Reasoning
LongBench v2
GLM-4.7—Kimi K2.561%SourceNot directly comparable
Knowledge
GPQA
GLM-4.785.7%SourceKimi K2.587.6%SourceKimi K2.5 leads this result
MMLU-Pro
GLM-4.784.3%SourceKimi K2.587.1%SourceKimi K2.5 leads this result
HLE
GLM-4.724.8%SourceKimi K2.530.1%SourceKimi K2.5 leads this result
GPQA-D
GLM-4.7—Kimi K2.587.6%SourceNot directly comparable
SuperGPQA
GLM-4.7—Kimi K2.569.2%SourceNot directly comparable
MMLU-Pro (Arcee)
GLM-4.7—Kimi K2.587.1%SourceNot directly comparable
Math
AIME 2025
GLM-4.795.7%SourceKimi K2.596.1%SourceKimi K2.5 leads this result
FrontierMath v2 (Tiers 1-3)
Shared sourceGLM-4.72.439%Kimi K2.527.900%Kimi K2.5 leads this result
FrontierMath v2 (Tier 4)
Shared sourceGLM-4.70.000%Kimi K2.54.200%Kimi K2.5 leads this result
AIME26
GLM-4.7—Kimi K2.595.8%SourceNot directly comparable
AIME25 (Arcee)
GLM-4.7—Kimi K2.596.3%SourceNot directly comparable
HMMT Feb 2025
GLM-4.7—Kimi K2.595.4%SourceNot directly comparable
HMMT Nov 2025
GLM-4.7—Kimi K2.591.1%SourceNot directly comparable
HMMT Feb 2026
GLM-4.7—Kimi K2.587.1%SourceNot directly comparable
MMAnswerBench
GLM-4.7—Kimi K2.581.8%SourceNot directly comparable
Multilingual
Multimodal
Instruction following
IFEval
GLM-4.7—Kimi K2.593.9%SourceNot directly comparable
Frequently asked questions
Which is better, GLM-4.7 or Kimi K2.5?
GLM-4.7 has the higher public score estimate, 60.91 versus 58.81, but the 90% score intervals overlap. The higher estimate is not a decisive winner because the uncertainty ranges overlap.
Which is better for coding, GLM-4.7 or Kimi K2.5?
The current coding averages use different weighted benchmark sets, so BenchLM does not name a winner from them. Read the shared benchmark rows directly and test the models on the same task set.
Which is better for agentic tasks, GLM-4.7 or Kimi K2.5?
Kimi K2.5 leads the like-for-like agentic tasks comparison across 2 shared weighted benchmark rows.
Which costs less, GLM-4.7 or Kimi K2.5?
Both models do not have comparable published API token rates, so this page does not name a universal price winner.
Which has the larger context window, GLM-4.7 or Kimi K2.5?
Kimi K2.5 has the larger documented context window: 256K, compared with 200K.
Related comparisons
Last updated September 3, 2026
Watch GLM-4.7 vs Kimi K2.5
One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.
Read a sample issueJoin 2,000+ readers.