Budget Picker

Cheap, Value, Mid, and Best are the top Terminal-Bench score in that list-price band. Quality is an editorial plan pick. Prices and scores refresh from the model table.

Budget / TaskModelWhy
Free codingNorth Mini CodeApache 2.0, FREE, TB 35.6, 3B active
Cheap agent (<$0.30/M out)DS V4 Flash$0.14/$0.28, TB 2.1 82.7, SWE-V 79.0
Value agent (<$1/M out)Hy3$0.13/$0.53, TB 71.7, Apache-2 open-weight
Open-weight agentMiniMax M3$0.30/$1.20, TB 66 (v), 1M ctx
Mid agent ($1-4/M out)GLM-5.2$0.76/$2.42, TB 81, SWE-Pro 62.1
Quality plan ($4-8/M out)GPT-5.6 Luna$1/$6, TB 84.7, SWE-Pro 62.7
Best agent (any budget)GPT-5.6 SolTB 88.8 SOTA, DeepSWE 72.7%, AA Coding #1

Recommended Model Pairs

Planner plus executor. Rating is editorial; names must exist in the model table.

RatingPlannerExecutorUse Case
★★Sonnet 5GLM-5.3Default agent pipeline (Coding Plan)
★★GPT-5.6 SolGLM-5.2Peak TB + MIT executor
GPT-5.6 LunaGLM-5.2Cost-efficient frontier
Sonnet 5Qwen3.8 MaxLong-horizon autonomous
Opus 4.8Qwen3.8 MaxPeak accuracy + long-horizon
Composer 2.5Grok Build 0.1Grok Build /goal mode
$Opus 4.8DS V4 FlashBudget code gen pipeline
$MiniMax M3Hy3Open-weight budget
FOSSGLM-5.2Qwen3.8-27BSelf-hosted agents

If you already pay for a provider

Best model on that catalog for agent/CLI work, cheap coding, and planning. Picks are derived from the live catalog and scores, not a frozen list.

ProviderAgent / CLIBudget codingPlanning
Z.AI DevPackGLM-5.3GLM-5.2GLM-5.2
ZenMuxGPT-5.6 SolDS V4 FlashClaude Fable 5
OpenCode GoDS V4 FlashDS V4 FlashDS V4 Pro
OpenRouterGrok 4.6Step 3.5FStep 3.5F
Ollama CloudDS V4 FlashDS V4 FlashDS V4 Pro
Nous Portal (via OpenRouter)GPT-5.6 SolDS V4 FlashClaude Fable 5
CrofAIKimi K3DS V4 FlashDS V4 Pro
FactoryGPT-5.6 SolDS V4 ProClaude Fable 5

If your provider lacks a model: OpenRouter bridges most providers | ZenMux has widest catalog | CrofAI is often the cheapest open-weight token source | Factory is a subscription coding agent