Model Follower.

Up to date

Rebuilt
2026-10-25 06:54 UTC
Release radar
117 on watch
Hub snapshot
2026-10-25
Re-checked
daily · every number sourced and dated

Use cases › Doing things

Pick a model for a coding agent

Write, fix and review code with an agent in the loop.

42 models with a coding score at these constraints, 21 with a third-party measurement for the job. Data observed 2026-10-25 06:54 UTC.

Showing every model that fits the job; narrow it with the constraints below.

What it looks like in practice

Start with these four

Each pick is the highest third-party SWE-bench Verified score that meets its constraint; no model appears twice. Author-reported scores stay in the table below.

Recipes

How people wire a model into this job. Notes are written by hand from the cited cards and docs; the model links go to the facts.

Kimi K3 in an OpenAI-compatible coding agent

Model
moonshotai/Kimi-K3
Harness
Any OpenAI-compatible agent (OpenCode, Aider, Claude Code via proxy)
Weights
Provider-hosted (native MXFP4 weights)
  • Point the agent at the provider's /v1 base URL and the provider's model ref from the where-to-run table.
  • Set max context per the provider row, not the config value; providers cap below the model maximum.
  • Measure cost per completed task on your own repo before committing; the model's own eval files are the lab's numbers, not ours.

Local coding agent on one GPU

Model
Qwen/Qwen3-Coder-30B-A3B-Instruct
Harness
vLLM or SGLang serving + Aider/OpenCode
Weights
FP8 (fits a single 80GB GPU with headroom; see hardware-fit estimate)
  • Architecture is registered in both vLLM and SGLang (see servability chips).
  • Enable tool calling in the server flags; check the model card's chat template notes.

Everything that fits 42

The first ten models with a coding score, in the scoreboard's order for this job. The full list, sortable, lives in the scoreboard with these constraints.

Everything that fits: first 10 of 42
#Evidence
1Kimi-K2.6moonshotai80.2 —$0.4518~616 GB
2Inkling-Smallthinkingmachines80.2 —$0.452~160 GB
3Inklingthinkingmachines77.6 —$0.954~571 GB
4Qwen3.6-27BQwen77.2 —$0.209~17 GB
5Muse-Glimmer-30Bmeta-models76 —$0.303~18 GB
6MiniMax-M2.5MiniMaxAI75.8 —$0.278~137 GB
7Step-3.5-Flashstepfun-ai74.4 —$0.102~120 GB
8MiniMax-M2.1MiniMaxAI74 —$0.302~137 GB
9GLM-4.7zai-org73.8 —$0.327~215 GB
10Qwen3.6-35B-A3BQwen73.4 —$0.05012~22 GB

Show all 42 in the scoreboard →

Most-shared on this job

Links about these models ranked by how many newsletters and link sources shared them in the last 28 days; each source counts once.All links

  1. Shared by 9
    GLM-5.3 and the spread of advanced cyber capabilitiesHacker News, Last Week in AI, TLDR AI, AI Breakfast and 5 moreAbout GLM-5
    anthropic.com · first seen 2026-09-29
  2. Shared by 1
    Microsoft Decision-1: 9B Qwen Model Launch ExplainedTLDR AIAbout Qwen3-Coder-480B-A35B-Instruct, Qwen3-Coder-30B-A3B-Instruct, Qwen3-235B-A22B-Instruct-2507 (series match)
    shattered.io · first seen 2026-10-23
  3. Shared by 1
    Early commenters dispute the published benchmark picture: one reports Flint is worse than Ornith-1.5-35B and “far worse…Latent Space / AINewsAbout Ornith-1.5-9B (series match)
    github.com · first seen 2026-10-23
  4. Shared by 1
    Early commenters dispute the published benchmark picture: one reports Flint is worse than Ornith-1.5-35B and “far worse…Latent Space / AINewsAbout Ornith-1.5-9B (series match)
    huggingface.co · first seen 2026-10-23
  5. Shared by 1
    Activity: 438): Ornith AI released Ornith-1.5-122B, a 122B MoE reasoning model fine-tuned/continued from Ornith-1.0 on…Latent Space / AINewsAbout Ornith-1.5-9B (series match)
    reddit.com · first seen 2026-10-22

How people judge this job

None of them measure how a model behaves in your own repository and harness. Scores depend on the agent scaffold and its settings, so compare within one benchmark, never across two.

Nearest scoreboard view: Coding agents. Follow changes to this job's evidence: RSS.