Agentic
Not comparable- Kimi K2.5
- 55.0
- Ministral 3 14B
- Not measured
- Weighted basis
- 2 vs 0 rows
- Reading
- Not comparable
Model comparison
Kimi K2.5 vs Ministral 3 14B
Updated July 29, 2026. Public scores include evidence status and uncertainty. They are not guarantees for a specific workload.
The public evidence has no benchmark result shared by both models, so it does not support a quality verdict. Use the documented cost, context, and runtime rows instead.
0 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.
Chat turn cost
1K fresh input + 500 output tokens
Ministral 3 14B
Ministral 3 14B has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.
Confidence: listed-rates
Cache-heavy agent loop cost
200K cached + 20K fresh input + 10K output tokens
Ministral 3 14B
Ministral 3 14B has the lower estimated token cost for this stated workload. Kimi K2.5 has no published cached-input rate, so cached tokens use its listed input rate. Ministral 3 14B has no published cached-input rate, so cached tokens use its listed input rate.
Confidence: rate-fallback
Show secondary and unsupported calls
Repository review cost
50K fresh input + 3K output tokens
Ministral 3 14B
Ministral 3 14B has the lower estimated token cost for this stated workload. Costs use the listed standard API rates.
Confidence: listed-rates
Coding work
Code generation, repair, and software-engineering tasks
Not enough matched evidence
No shared weighted benchmark basis supports a winner.
Confidence: limited
Agentic work
Tool use, computer use, and multi-step task completion
Not enough matched evidence
No shared weighted benchmark basis supports a winner.
Confidence: limited
Long documents
Prompts that approach the documented context limit
No clear pick
The documented context windows are equal.
Confidence: documented
What is actually comparable
Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.
45 Kimi K2.5 only
- Shared results
- 0
- Kimi K2.5 only
- 45
- Ministral 3 14B only
- 0
- Like-for-like categories
- 0 / 8
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
Not comparable- Kimi K2.5
- 59.4
- Ministral 3 14B
- Not measured
- Weighted basis
- 4 vs 0 rows
- Reading
- Not comparable
Reasoning
Not comparable- Kimi K2.5
- 61.0
- Ministral 3 14B
- Not measured
- Weighted basis
- 1 vs 0 rows
- Reading
- Not comparable
Knowledge
Not comparable- Kimi K2.5
- 56.9
- Ministral 3 14B
- Not measured
- Weighted basis
- 4 vs 0 rows
- Reading
- Not comparable
Math
Not comparable- Kimi K2.5
- 60.6
- Ministral 3 14B
- Not measured
- Weighted basis
- 4 vs 0 rows
- Reading
- Not comparable
Multilingual
Not comparable- Kimi K2.5
- 82.3
- Ministral 3 14B
- Not measured
- Weighted basis
- 1 vs 0 rows
- Reading
- Not comparable
Multimodal
Not comparable- Kimi K2.5
- 78.5
- Ministral 3 14B
- Not measured
- Weighted basis
- 1 vs 0 rows
- Reading
- Not comparable
Instruction following
Not comparable- Kimi K2.5
- 93.9
- Ministral 3 14B
- Not measured
- Weighted basis
- 1 vs 0 rows
- Reading
- Not comparable
| Category | Kimi K2.5 | Ministral 3 14B | Weighted basis | Reading |
|---|---|---|---|---|
| Agentic | 55.0 | Not measured | Not comparable2 vs 0 rows | Not comparable |
| Coding | 59.4 | Not measured | Not comparable4 vs 0 rows | Not comparable |
| Reasoning | 61.0 | Not measured | Not comparable1 vs 0 rows | Not comparable |
| Knowledge | 56.9 | Not measured | Not comparable4 vs 0 rows | Not comparable |
| Math | 60.6 | Not measured | Not comparable4 vs 0 rows | Not comparable |
| Multilingual | 82.3 | Not measured | Not comparable1 vs 0 rows | Not comparable |
| Multimodal | 78.5 | Not measured | Not comparable1 vs 0 rows | Not comparable |
| Instruction following | 93.9 | Not measured | Not comparable1 vs 0 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.
A shared-evidence shape is not available.
BenchLM does not draw a radar or infer missing axes when the matched evidence is too sparse.
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
- Kimi K2.5
- $0.0021
- Fits in one request
- Ministral 3 14B
- $0.0003
- Fits in one request
Ministral 3 14B has the lower modeled cost
Costs use the listed standard API rates.
Repository review
50K fresh input + 3K output tokens
- Kimi K2.5
- $0.039
- Fits in one request
- Ministral 3 14B
- $0.0106
- Fits in one request
Ministral 3 14B has the lower modeled cost
Costs use the listed standard API rates.
Cache-heavy agent loop
200K cached + 20K fresh input + 10K output tokens
- Kimi K2.5
- $0.162
- Fits in one request
- Cached input priced at the published list-input rate
- Ministral 3 14B
- $0.046
- Fits in one request
- Cached input priced at the published list-input rate
Ministral 3 14B has the lower modeled cost
Kimi K2.5 has no published cached-input rate, so cached tokens use its listed input rate. Ministral 3 14B has no published cached-input rate, so cached tokens use its listed input rate.
Specification differences
Sourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.
SpecificationKimi K2.5Ministral 3 14B
Context window
Maximum documented context; output-token limits may be lower.
Kimi K2.5
256K
Ministral 3 14B
256K
API model ID
Kimi K2.5
Not sourced
Ministral 3 14B
Not sourced
Cached-input rate
A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.
Kimi K2.5
Not published
Ministral 3 14B
Not published
Documented inputs
Kimi K2.5
Not sourced
Ministral 3 14B
Not sourced
Documented outputs
Kimi K2.5
Not sourced
Ministral 3 14B
Not sourced
Provider availability
Kimi K2.5
Not sourced
Ministral 3 14B
Not sourced
Reasoning profile
Kimi K2.5
Non-Reasoning
Ministral 3 14B
Non-Reasoning
Weight access
Kimi K2.5
Open Weight
Ministral 3 14B
Open Weight
License
Kimi K2.5
Open Weight
Ministral 3 14B
Open Weight
Release date
Kimi K2.5
2026-02-01
Ministral 3 14B
2025-12-02
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
- The public evidence has no benchmark result shared by both models, so it does not support a quality verdict.
- Workload cost
- Repository review: $0.039 vs $0.0106. Cache-heavy agent loop: $0.162 vs $0.046.
- Context tradeoff
- Both models list 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.
Kimi K2.5
API / mo$2,700
Self-host / mo$5,221
Break-even132M/day
Ministral 3 14B
API / mo$300
Self-host / moNot listed
Break-even—
Proprietary model — self-hosting not applicable.
Benchmark evidence
The full public result ledger is available for audit without forcing a wide desktop table onto a phone.
Browse raw public benchmark evidence45 rows
Agentic
Terminal-Bench 2.0
Kimi K2.550.8%SourceMinistral 3 14B—Not directly comparable
BrowseComp
Kimi K2.560.6%SourceMinistral 3 14B—Not directly comparable
Claw-Eval
Kimi K2.552.3%SourceMinistral 3 14B—Not directly comparable
QwenClawBench
Kimi K2.554.3%SourceMinistral 3 14B—Not directly comparable
τ³-bench results
Kimi K2.565.7%SourceMinistral 3 14B—Not directly comparable
DeepSearchQA
Kimi K2.577.1%SourceMinistral 3 14B—Not directly comparable
DeepPlanning
Kimi K2.514.4%SourceMinistral 3 14B—Not directly comparable
Toolathlon
Kimi K2.527.8%SourceMinistral 3 14B—Not directly comparable
MCP Atlas
Kimi K2.529.5%SourceMinistral 3 14B—Not directly comparable
MCP-Tasks
Kimi K2.559.1%SourceMinistral 3 14B—Not directly comparable
WideResearch
Kimi K2.572.7%SourceMinistral 3 14B—Not directly comparable
Gert Labs
Kimi K2.545.88%SourceMinistral 3 14B—Not directly comparable
ResearchClawBench
Kimi K2.514.0%SourceMinistral 3 14B—Not directly comparable
JobBench
Kimi K2.58.7%SourceMinistral 3 14B—Not directly comparable
Coding
SWE-bench Verified
Kimi K2.576.8%SourceMinistral 3 14B—Not directly comparable
SWE-bench Verified*
Kimi K2.570.8%SourceMinistral 3 14B—Not directly comparable
LiveCodeBench v6
Kimi K2.585.0%SourceMinistral 3 14B—Not directly comparable
SWE-bench Pro
Kimi K2.550.7%SourceMinistral 3 14B—Not directly comparable
SWE Multilingual
Kimi K2.573%SourceMinistral 3 14B—Not directly comparable
SWE-Rebench
Kimi K2.558.5%SourceMinistral 3 14B—Not directly comparable
React Native Evals
Kimi K2.577.2%SourceMinistral 3 14B—Not directly comparable
SciCode
Kimi K2.548.7%SourceMinistral 3 14B—Not directly comparable
Reasoning
LongBench v2
Kimi K2.561%SourceMinistral 3 14B—Not directly comparable
Knowledge
GPQA
Kimi K2.587.6%SourceMinistral 3 14B—Not directly comparable
GPQA-D
Kimi K2.587.6%SourceMinistral 3 14B—Not directly comparable
SuperGPQA
Kimi K2.569.2%SourceMinistral 3 14B—Not directly comparable
MMLU-Pro
Kimi K2.587.1%SourceMinistral 3 14B—Not directly comparable
MMLU-Pro (Arcee)
Kimi K2.587.1%SourceMinistral 3 14B—Not directly comparable
HLE
Kimi K2.530.1%SourceMinistral 3 14B—Not directly comparable
Math
AIME 2025
Kimi K2.596.1%SourceMinistral 3 14B—Not directly comparable
AIME26
Kimi K2.595.8%SourceMinistral 3 14B—Not directly comparable
AIME25 (Arcee)
Kimi K2.596.3%SourceMinistral 3 14B—Not directly comparable
HMMT Feb 2025
Kimi K2.595.4%SourceMinistral 3 14B—Not directly comparable
HMMT Nov 2025
Kimi K2.591.1%SourceMinistral 3 14B—Not directly comparable
HMMT Feb 2026
Kimi K2.587.1%SourceMinistral 3 14B—Not directly comparable
MMAnswerBench
Kimi K2.581.8%SourceMinistral 3 14B—Not directly comparable
FrontierMath v2 (Tiers 1-3)
Kimi K2.527.900%SourceMinistral 3 14B—Not directly comparable
FrontierMath v2 (Tier 4)
Kimi K2.54.200%SourceMinistral 3 14B—Not directly comparable
Multilingual
Multimodal
Instruction following
IFEval
Kimi K2.593.9%SourceMinistral 3 14B—Not directly comparable
Frequently asked questions
Which is better, Kimi K2.5 or Ministral 3 14B?
The public evidence has no benchmark result shared by both models, so it does not support a quality verdict. The page therefore keeps the decision tied to the specific documented workload.
Which is better for coding, Kimi K2.5 or Ministral 3 14B?
The published evidence does not provide a shared weighted coding basis for both models, so BenchLM does not name a coding winner.
Which is better for agentic tasks, Kimi K2.5 or Ministral 3 14B?
The published evidence does not provide a shared weighted agentic tasks basis for both models, so BenchLM does not name a agentic tasks winner.
Which costs less, Kimi K2.5 or Ministral 3 14B?
For the stated presets, chat costs $0.0021 on Kimi K2.5 and $0.0003 on Ministral 3 14B; repository review costs $0.039 and $0.0106; the cache-heavy agent loop costs $0.162 and $0.046. Kimi K2.5 has no published cached-input rate, so cached tokens use its listed input rate. Ministral 3 14B has no published cached-input rate, so cached tokens use its listed input rate.
Which has the larger context window, Kimi K2.5 or Ministral 3 14B?
Both models list the same context window, 256K.
Related comparisons
Last updated July 29, 2026
Watch Kimi K2.5 vs Ministral 3 14B
One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.
Read a sample issueJoin 2,000+ readers.