The LLM market is too crowded to pick by brand memory. This page is a decision tool: every card says when to use a model, when to skip it and which data backs that call. Filter by use case, price and market position.
Updated 2026-06-14 · Next review by 2026-06-28 · What changed →
Three quick choices without scrolling: one for the hardest work, one for cheap volume and one for teams that do not want to depend only on a closed API.
My read
coding pick backed by current AA.
Claude Fable 5 has IQ 64.9 and input $10/M in the sources. Consider it for coding, AI agents; for mass-volume, real-time-latency, run a second benchmark before rollout.
premium, when mistakes hurt · market frontier · verified by external data
When to pick it →My read
coding pick backed by current AA.
GPT-5.5 has IQ 60.2 and input $5/M and DeepSWE pass@1 70.0 % in the sources. Consider it for coding, AI agents; for self-hosted stack, real-time-latency, run a second benchmark before rollout.
mid-budget · market frontier · verified by external data
When to pick it →My read
batch pick backed by current AA.
DeepSeek V4 Flash has IQ 46.5 and input $0.14/M in the sources. Consider it for large batches, fast replies; for deep-frontier-reasoning, top-coding, run a second benchmark before rollout.
open / self-run · specialist · verified by external data
When to pick it →This is the broader catalogue. Filters narrow it by situation; details keep hard numbers and sources out of the first read.
My read
coding pick backed by current AA.
Claude Fable 5 has IQ 64.9 and input $10/M in the sources. Consider it for coding, AI agents; for mass-volume, real-time-latency, run a second benchmark before rollout.
premium, when mistakes hurt · market frontier · verified by external data
When to pick it →My read
coding pick backed by current AA.
Claude Opus 4.8 has IQ 61.4 and input $5/M and DeepSWE pass@1 58.2 % in the sources. Consider it for coding, AI agents; for mass-volume, real-time-latency, run a second benchmark before rollout.
mid-budget · market frontier · verified by external data
When to pick it →My read
coding pick backed by current AA.
GPT-5.5 has IQ 60.2 and input $5/M and DeepSWE pass@1 70.0 % in the sources. Consider it for coding, AI agents; for self-hosted stack, real-time-latency, run a second benchmark before rollout.
mid-budget · market frontier · verified by external data
When to pick it →My read
rag pick backed by current AA.
Gemini 3.1 Pro Preview has IQ 57.2 and input $2/M and DeepSWE pass@1 9.7 % in the sources. Consider it for company knowledge, vision and multimodal; for real-time-latency, self-hosted stack, run a second benchmark before rollout.
mid-budget · market frontier · verified by external data
When to pick it →My read
multilingual pick backed by current AA.
Qwen3.7 Max has IQ 56.6 and input $2.5/M and DeepSWE pass@1 17.7 % in the sources. Consider it for multilingual content, coding; for self-hosted stack, premium-agents, run a second benchmark before rollout.
mid-budget · market frontier · verified by external data
When to pick it →My read
batch pick backed by current AA.
Gemini 3.5 Flash has IQ 55.3 and input $1.5/M and DeepSWE pass@1 28.3 % in the sources. Consider it for large batches, company knowledge; for deep-coding, real-time-latency, run a second benchmark before rollout.
mid-budget · specialist · verified by external data
When to pick it →My read
batch pick backed by current AA.
Kimi K2.6 has IQ 53.9 and input $0.95/M and DeepSWE pass@1 23.9 % in the sources. Consider it for large batches, coding; for enterprise-governance, tool-use, run a second benchmark before rollout.
cheap at volume · specialist · verified by external data
When to pick it →My read
coding pick backed by current AA.
Claude Sonnet 4.6 has IQ 51.7 and input $3/M and DeepSWE pass@1 31.8 % in the sources. Consider it for coding, AI agents; for self-hosted stack, mass-volume, run a second benchmark before rollout.
mid-budget · specialist · verified by external data
When to pick it →My read
self-hosted pick backed by current AA.
GLM-5.1 has IQ 51.4 and input $1.4/M and DeepSWE pass@1 17.5 % in the sources. Consider it for self-hosted stack, sensitive deployments; for premium-agents, top-coding, run a second benchmark before rollout.
open / self-run · specialist · verified by external data
When to pick it →My read
batch pick backed by current AA.
DeepSeek V4 Flash has IQ 46.5 and input $0.14/M in the sources. Consider it for large batches, fast replies; for deep-frontier-reasoning, top-coding, run a second benchmark before rollout.
open / self-run · specialist · verified by external data
When to pick it →My read
batch pick backed by current AA.
DeepSeek V4 Pro has IQ 51.5 and input $0.435/M and DeepSWE pass@1 7.5 % in the sources. Consider it for large batches, coding; for enterprise-governance, premium-agents, run a second benchmark before rollout.
open / self-run · specialist · verified by external data
When to pick it →My read
rag pick backed by current AA.
Command A Plus has IQ 37.2 and input $0/M in the sources. Consider it for company knowledge, document extraction; for top-coding, deep-frontier-reasoning, run a second benchmark before rollout.
cheap at volume · specialist · verified by external data
When to pick it →My read
coding pick backed by current AA.
Grok 4.3 has IQ 53.2 and input $1.25/M in the sources. Consider it for coding, fast replies; for self-hosted stack, sensitive deployments, run a second benchmark before rollout.
mid-budget · specialist · verified by external data
When to pick it →My read
self-hosted pick backed by current AA.
Llama 4 Maverick has IQ 18.4 and input $0.35/M in the sources. Consider it for self-hosted stack, sensitive deployments; for managed-api-comfort, deep-frontier-reasoning, run a second benchmark before rollout.
open / self-run · specialist · verified by external data
When to pick it →My read
compliance pick backed by current AA.
Mistral Medium 3.5 has IQ 39.2 and input $1.5/M in the sources. Consider it for sensitive deployments, company knowledge; for deep-frontier-reasoning, top-coding, run a second benchmark before rollout.
mid-budget · specialist · verified by external data
When to pick it →My read
compliance pick backed by current AA.
Mistral Large 3 has IQ 22.8 and input $0.5/M in the sources. Consider it for sensitive deployments, document extraction; for deep-frontier-reasoning, AI agents, run a second benchmark before rollout.
cheap at volume · specialist · verified by external data
When to pick it →Curated decision snapshot, not a live leaderboard. Primary data comes from Artificial Analysis, with LMArena, LLM Stats, Aider, SWE-Bench, DeepSWE and HF Open LLM Leaderboard used when available during the run. The page is reviewed roughly every two weeks and cards without data show a loading state instead of invented numbers.
