Hugging Face's logo

Buckets:
celinah
/
hf-cli-agent-benchmark

Files
xet
celinah/hf-cli-agent-benchmark / T07-find-coding-models /claude-code-sonnet /cli-no-skill /cli_no_skill-r0-59fc7c.jsonl
celinah's picture
celinah
trace code
|
download
raw
48.3 kB
Claude Code trace
User 2026-06-01T12:00:00.000Z

Find up to 8 trending text-generation models for coding, released in 2025-2026, with fewer than 70B parameters and suitable for local deployment. Exclude the mainstream Llama, Mistral and Qwen base series. Return the model IDs with their parameter counts.

Assistant claude-code-sonnet 2026-06-01T12:00:02.000Z

I'll search for trending coding-focused text-generation models on Hugging Face using the CLI.

Assistant claude-code-sonnet 2026-06-01T12:00:03.000Z
2 tool calls (Bash)
Assistant claude-code-sonnet 2026-06-01T12:00:08.000Z

Now I'll search for trending coding models with different queries.

Assistant claude-code-sonnet 2026-06-01T12:00:09.000Z
1 tool call (Bash)
Assistant claude-code-sonnet 2026-06-01T12:00:12.000Z
1 tool call (Bash)
Assistant claude-code-sonnet 2026-06-01T12:00:15.000Z

Let me search with more targeted queries for specific coding model families from 2025-2026.

Assistant claude-code-sonnet 2026-06-01T12:00:16.000Z
3 tool calls (Bash)
Assistant claude-code-sonnet 2026-06-01T12:00:23.000Z
1 tool call (mcp__claude_ai_Hugging_Face__hub_repo_details)
Assistant claude-code-sonnet 2026-06-01T12:00:26.000Z
1 tool call (mcp__claude_ai_Hugging_Face__hub_repo_details)
Assistant claude-code-sonnet 2026-06-01T12:00:29.000Z
2 tool calls (mcp__claude_ai_Hugging_Face__hub_repo_details, Bash)
Assistant claude-code-sonnet 2026-06-01T12:00:34.000Z
1 tool call (Bash)
Assistant claude-code-sonnet 2026-06-01T12:00:37.000Z

Let me check a few more specific models to finalize my selection.

Assistant claude-code-sonnet 2026-06-01T12:00:38.000Z
1 tool call (mcp__claude_ai_Hugging_Face__hub_repo_details)
Assistant claude-code-sonnet 2026-06-01T12:00:41.000Z

I now have enough data to compile the final list. Here are 8 trending text-generation models for coding, released in 2025–2026, under 70B parameters, suitable for local deployment, excluding mainstream Llama/Mistral/Qwen base series:

Model Params Released Notes
JetBrains/Mellum-4b-base ~4B May 2025 Dedicated code model
JetBrains/Mellum2-12B-A2.5B-Thinking ~12.15B (MoE, 2.5B active) Jun 2026 Code-focused thinking model
microsoft/Phi-4-mini-instruct ~3.84B Mar 2025 Strong at code, MIT license
microsoft/Phi-4-reasoning-plus ~14.66B Apr 2025 Code + math reasoning
ibm-granite/granite-3.3-8b-instruct ~8.17B May 2025 IBM Granite, code capable
ibm-granite/granite-4.0-tiny-preview ~6.67B (MoE) Nov 2025 IBM Granite 4.0 series
Tesslate/OmniCoder-9B ~9.41B Mar 2026 Dedicated coding/agent model
nvidia/OpenReasoning-Nemotron-32B ~32.76B Sep 2025 Code + STEM reasoning

TASK_COMPLETE

json 
{
  "models": [
    {"model_id": "JetBrains/Mellum-4b-base", "parameters": "~4B"},
    {"model_id": "JetBrains/Mellum2-12B-A2.5B-Thinking", "parameters": "~12.15B (MoE, 2.5B active)"},
    {"model_id": "microsoft/Phi-4-mini-instruct", "parameters": "~3.84B"},
    {"model_id": "microsoft/Phi-4-reasoning-plus", "parameters": "~14.66B"},
    {"model_id": "ibm-granite/granite-3.3-8b-instruct", "parameters": "~8.17B"},
    {"model_id": "ibm-granite/granite-4.0-tiny-preview", "parameters": "~6.67B (MoE)"},
    {"model_id": "Tesslate/OmniCoder-9B", "parameters": "~9.41B"},
    {"model_id": "nvidia/OpenReasoning-Nemotron-32B", "parameters": "~32.76B"}
  ]
}