Qwen 2.5 Coder 7B
A Q6_K-GGUF distribution of Qwen 2.5 Coder 7B for local coding: 7.6 billion dense parameters, Apache-2.0 license, specialized in code generation, debugging, and repair. The family supports 128,000 tokens of context; in GGUF setups, only 32,000 are natively available without long-context configuration. Fully commercially usable, compact and efficient on Workstation hardware.
- Open Weights
- Edge
- llama.cpp
- Text
- Instruction-Tuned
- Interactive
Sovereign Risk: LOW The weights originate from Alibaba’s Apache-2.0-licensed Qwen2.5-Coder family and are run entirely locally here. Without a cloud connection, operational risk is low; the provenance remains Chinese-jurisdictional, but local Open Weights usage minimizes data exposure.[web:875][web:876][web:878]
Key metrics
Score · Latency · Cost · Quality
- Total Score Bronze
- 55.32
- Routine
- 33.87
- Reasoning
- 21.45
- LLM Judge Avg
- 2.53 / 5
- 100 Coverage
- Avg Task Duration
- 33.7s
- Interactive
- Token Rate
- 37.43tok/s
- Output Rate
- P95 Latency
- 46.54s
- Top 5 %
- Total Tokens
- 68800
- Output Volume
- Cost per 1K
- $0
- USD / 1K Requests
- Benchmark Cost
- $0
- Total · 68800 tok
Benchmark modules
10 modules · weighted · vs. model median & top performer
Token efficiency & latency
Consumption per module vs. model median
Token consumption per module
Performance profile
Key metrics – legend
Score
- Total Score
- Overall score derived from Routine and Reasoning scores. Basis for the fleet rank and tier assignment (Gold / Silver / Bronze / Standard).
- Routine
- Score for standard tasks: code generation, text transformation, structured outputs. Weighted by task complexity.
- Reasoning
- Score for analytical and logical tasks: multi-step problem solving, mathematical reasoning, structured argumentation.
- Rank
- Position in the fleet comparison, sorted by Total Score. Refers to the snapshot at the time of testing.
LLM Judge
- LLM Judge Ø
- Average quality rating by an independent judge model on a scale of 1–5.
- Coverage
- Share of tasks for which a judge rating is available. 100% means all responses were evaluated.
Latency & speed
- Avg. Task Duration
- Average time per task from the first to the last token of the response, measured across all benchmark tasks.
- Token Rate
- Output speed in tokens per second. Relevant for interactive applications and real-time streaming.
- P95 Latency
- 95th percentile of response times — shows outliers. Lower values mean fewer long wait times in edge cases.
Cost
- Cost per 1K
- API price per 1,000 output tokens in US dollars. Basis for cost calculation in production use.
- Benchmark Cost
- Total cost for the complete benchmark run of this model.