Medium (14–34B) AI models

All 114 open medium models. The single-GPU sweet spot — strong and self-hostable. Compare calculated local requirements, provider-published rented-GPU rates, and dated API rates or clearly labeled estimates.

• Qwen2.5-Coder 32B Instruct — 32B, Alibaba · $0.83/1M API
• gpt-oss-20b — 21B, OpenAI · $0.08/1M API
• DeepSeek-R1-Distill-Qwen-32B — 32.5B, DeepSeek · $0.36/1M API est.
• Qwen3-32B — 32.8B, Alibaba · $0.18/1M API
• Gemma 3 27B — 27B, Google · $0.27/1M API
• Phi-4 — 14B, Microsoft · $0.11/1M API
• Gemma 2 27B Instruct — 27B, Google · $0.65/1M API
• Mistral Small 3 (24B, 2501) — 23.6B, Mistral AI · $0.07/1M API
• Qwen2.5 Coder 14B Instruct — 14.8B, Qwen · $0.22/1M API est.
• Qwen3 30B A3B — 30.5B, Qwen · $0.31/1M API
• Qwen2.5 14B Instruct — 14.8B, Qwen · $0.22/1M API est.
• Qwen3.6 35B A3B NVFP4 — 18.7B, nvidia · $0.25/1M API est.
• Qwen3 14B — 14.8B, Qwen · $0.18/1M API
• Gemma 4 31B IT NVFP4 — 20.9B, nvidia · $0.27/1M API est.
• Qwen3 Coder 30B A3B Instruct — 30.5B, Qwen · $0.18/1M API
• GLM 4.7 Flash — 31.2B, zai-org · $0.23/1M API
• NVIDIA Nemotron 3 Nano 30B A3B BF16 — 31.6B, nvidia · $0.35/1M API est.
• DeepSeek-Coder-V2-Lite Instruct — 15.7B, DeepSeek · $0.23/1M API est.
• Gemma 4 26B A4B NVFP4 — 14.4B, nvidia · $0.22/1M API est.
• OTel 2.0 LLM 31B IT — 32.1B, farbodtavakkoli · $0.36/1M API est.
• DeepSeek V2 Lite Chat — 15.7B, deepseek-ai · $0.23/1M API est.
• Qwen2.5 32B Instruct — 32.8B, Qwen · $0.36/1M API est.
• Qwen3 30B A3B Instruct 2507 — 30.5B, Qwen · $0.12/1M API
• NVIDIA Nemotron 3 Nano 30B A3B NVFP4 — 18.2B, nvidia · $0.25/1M API est.
• gpt neox 20b — 20.7B, EleutherAI · $0.27/1M API est.
• Qwen3.6 27B NVFP4 — 18.2B, nvidia · $0.25/1M API est.
• granite 4.0 h small — 32.2B, ibm-granite · $0.36/1M API est.
• DeepSeek R1 Distill Qwen 14B — 14.8B, deepseek-ai · $0.22/1M API est.
• Qwen3.6 27B Text NVFP4 MTP — 16.7B, sakamakismile · $0.23/1M API est.
• Qwen3.8 27B OBLITERATED — 27.8B, OBLITERATUS · $0.32/1M API est.
• Qwen3 30B A3B abliterated — 30.5B, mlabonne · $0.34/1M API est.
• GLM 4.7 Flash — 31.2B, unsloth · $0.35/1M API est.
• DeepSeek V2 Lite — 15.7B, deepseek-ai · $0.23/1M API est.
• droplychee 1.0 27b — 27.8B, droplychee · $0.32/1M API est.
• Laguna XS.2 — 33.4B, poolside · $0.15/1M API last-known
• NVIDIA Nemotron 3.5 Lightning 30B A3B NVFP4 — 17.8B, nvidia · $0.24/1M API est.
• diffusiongemma 26B A4B it NVFP4 — 14.4B, nvidia · $0.22/1M API est.
• gemma 4 26B A4B it uncensored — 25.8B, TrevorJS · $0.31/1M API est.
• GLM 4.7 Flash NVFP4 — 18.4B, GadflyII · $0.25/1M API est.
• Ornith 1.5 35B A3B NVFP4 — 18.7B, ornith-ai · $0.25/1M API est.
• Qwen3 30B A3B Thinking 2507 — 30.5B, Qwen · $1.30/1M API
• LLaDA2.0 mini — 16.3B, inclusionAI · $0.23/1M API est.
• gemma 4 31B it NVFP4 turbo — 32.5B, LilaRest · $0.36/1M API est.
• Nemotron 3 Nano 30B A3B — 31.6B, unsloth · $0.35/1M API est.
• Qwen1.5 MoE A2.7B — 14.3B, Qwen · $0.21/1M API est.
• Gemma 4 26B A4B it NVFP4 — 15.1B, bg-digitalservices · $0.22/1M API est.
• Agents A1 NVFP4 — 18.9B, r0b0tlab · $0.25/1M API est.
• Tongyi DeepResearch 30B A3B — 30.5B, Alibaba-NLP · $0.34/1M API est.
• Qwen2.5 14B Instruct — 14.8B, unsloth · $0.22/1M API est.
• phi 4 quantized.w4a16 — 14.8B, RedHatAI · $0.22/1M API est.
• Gemma 4 Garnet V2 31B it ultra uncensored heretic — 31.3B, llmfan46 · $0.35/1M API est.
• Qwen3.6 27B NVFP4 — 17.1B, ocicek · $0.24/1M API est.
• Qwen3 30B A3B Base — 30.5B, Qwen · $0.34/1M API est.
• LLaDA2.1 mini — 16.3B, inclusionAI · $0.23/1M API est.
• HyperCLOVAX SEED Think 32B — 33.3B, naver-hyperclovax · $0.37/1M API est.
• EuroLLM 22B Instruct 2512 — 22.6B, utter-project · $0.28/1M API est.
• lynx instruct 30b — 30.5B, bineric · $0.34/1M API est.
• NVIDIA Nemotron 3 Nano 30B A3B Base BF16 — 31.6B, nvidia · $0.35/1M API est.
• Qwen3 30B A3B NVFP4 — 15.6B, nvidia · $0.22/1M API est.
• Gemma 4 26B A4B it Uncensored NVFP4 — 15.1B, AEON-7 · $0.22/1M API est.
• Qwen2.5 14B — 14.8B, Qwen · $0.22/1M API est.
• cogito v1 preview qwen 32B — 32.8B, deepcogito · $0.36/1M API est.
• Qwen3 30B A3B NVFP4 — 17.5B, RedHatAI · $0.24/1M API est.
• Huihui Qwen3.6 27B abliterated NVFP4 MTP — 17.1B, sakamakismile · $0.24/1M API est.
• granite 4.1 30b — 28.9B, ibm-granite · $0.33/1M API est.
• Qwen2.5 Coder 14B — 14.8B, Qwen · $0.22/1M API est.
• NVIDIA Nemotron 3.5 Lightning 30B A3B BF16 — 31.6B, nvidia · $0.35/1M API est.
• Qwen3 32B NVFP4 — 17.2B, nvidia · $0.24/1M API est.
• Qwen3.6 27B AEON Ultimate Uncensored NVFP4 — 19.1B, AEON-7 · $0.25/1M API est.
• Qwen3 14B Base — 14.8B, Qwen · $0.22/1M API est.
• gpt oss 20b BF16 — 20.9B, unsloth · $0.27/1M API est.
• Qwen3.6 27B Claude Opus Sonnet Distilled NVFP4 MTP — 19.6B, Brian6145 · $0.26/1M API est.
• Qwen2.5 32B — 32.8B, Qwen · $0.36/1M API est.
• Qwen3 14B Instruct — 14.8B, OpenPipe · $0.22/1M API est.
• Olmo 3 1125 32B — 32.2B, allenai · $0.36/1M API est.
• Moonlight 16B A3B Instruct — 16B, moonshotai · $0.23/1M API est.
• QwQ 32B — 32.8B, Qwen · $0.36/1M API est.
• sarvam 30b — 32.2B, sarvamai · $0.36/1M API est.
• medgemma 27b text it — 27B, google · $0.32/1M API est.
• deepseek moe 16b base — 16.4B, deepseek-ai · $0.23/1M API est.
• Qwen3.6 27B AEON Ultimate Uncensored Multimodal NVFP4 MTP — 19.6B, AEON-7 · $0.26/1M API est.
• solar pro preview instruct — 22.1B, upstage · $0.28/1M API est.
• Qwen3.6 27B OBLITERATED — 26.9B, OBLITERATUS · $0.32/1M API est.
• Qwen3.6 27B AEON Ultimate Uncensored Multimodal NVFP4 MTP XS — 17.1B, AEON-7 · $0.24/1M API est.
• gpt oss safeguard 20b — 21.5B, openai · $0.19/1M API
• Param2 17B A2.4B Thinking — 17.2B, bharatgenai · $0.24/1M API est.
• Nemotron Cascade 2 30B A3B — 31.6B, nvidia · $0.35/1M API est.
• Qwen3.8 27B MTP NVFP4 — 27.4B, sakamakismile · $0.32/1M API est.
• Qwen3.8 27B AEON ULTIMATE UNCENSORED NVFP4 — 27.4B, sakamakismile · $0.32/1M API est.
• LFM2 24B A2B — 23.8B, LiquidAI · $0.29/1M API est.
• Ling mini 2.0 — 16.3B, inclusionAI · $0.23/1M API est.
• Qwen3.6 27B AEON Ultimate Uncensored BF16 — 27.4B, AEON-7 · $0.32/1M API est.
• Huihui Qwen3.8 27B abliterated NVFP4 — 27.4B, sakamakismile · $0.32/1M API est.
• starcoder — 15.8B, bigcode · $0.23/1M API est.
• HyperCLOVAX SEED Think 14B — 14.7B, naver-hyperclovax · $0.22/1M API est.
• Qwen3.6 27B DSV4Pro Thinking Distill — 26.9B, nerkyor · $0.32/1M API est.
• Ornith 1.0 35B AEON Ultimate Uncensored NVFP4 — 21B, AEON-7 · $0.27/1M API est.
• Qwopus3.6 27B Coder — 27.8B, Jackrong · $0.32/1M API est.
• Laguna XS 2.1 NVFP4 — 33.4B, poolside · $0.37/1M API est.
• llm jp 4 32b a3b thinking — 32.1B, llm-jp · $0.36/1M API est.
• ERNIE 4.5 21B A3B Thinking — 21.8B, baidu · $0.27/1M API est.
• Fastino Nemotron 3.5 Lightning Healthcare — 31.6B, fastino · $0.35/1M API est.
• Qwen3.8 27B AEON ULTIMATE UNCENSORED BF16 — 27.8B, AEON-7 · $0.32/1M API est.
• Qwen3.6 35B A3B abliterated NVFP4 MTP — 20.4B, THe-Plague · $0.26/1M API est.
• North Mini Code 1.0 — 30.5B, CohereLabs · $0.34/1M API est.
• Trinity Mini — 26.1B, arcee-ai · $0.10/1M API last-known
• Qwen3 Coder 30B A3B Instruct FP4 — 15.6B, NVFP4 · $0.22/1M API est.
• deepseek moe 16b chat — 16.4B, deepseek-ai · $0.23/1M API est.
• qwen27B Agent R2 abliterated preview — 26.9B, hotdogs · $0.32/1M API est.
• Dolphin Mistral 24B Venice Edition — 24B, dphn · $0.29/1M API est.
• Laguna XS 2.1 — 33.4B, poolside · $0.09/1M API
• Ornith 1.0 35B PrismaAURA 4.75bit vllm MTP — 21.9B, rdtand · $0.28/1M API est.
• Moonlight 16B A3B — 16B, moonshotai · $0.23/1M API est.
• Phi 4 reasoning plus — 14.7B, microsoft · $0.22/1M API est.

Other sizes

Flagship (80B+) · Large (34–80B) · Small (4–14B) · Tiny (under 4B) · All models

Compare → · LLM cost calculator →

Open the free LLM cost calculator → · Rate table dated 2026-08-24. $0 markup, your own accounts, we never resell compute. © 2026 Cynosure LLC.