--- title: "Is Llama 3.2 3B Instruct still undominated? · Undominated.ai" canonical: https://undominated.ai/check/meta-llama__llama-3.2-3b-instruct/ description: "Llama 3.2 3B Instruct: whether it is still undominated. If anything in this catalogue is both better and cheaper, it is named, as of Sep 23, 2026. Unrated stays unrated." --- # Is Llama 3.2 3B Instruct still undominated? · Undominated.ai > Llama 3.2 3B Instruct: whether it is still undominated. If anything in this catalogue is both better and cheaper, it is named, as of Sep 23, 2026. Unrated stays unrated. # Is Llama 3.2 3B Instruct a good deal? Whether anything in this catalogue beats Llama 3.2 3B Instruct on both quality and price, and what you give up if it does. A computation on the current catalogue, not an opinion. 13 undominated of 136 · Sep 23, 2026 [1 · Choose & set workload](#check-input)[2 · Read verdict](#check-verdict)[3 · Inspect alternatives](#check-options) As of Sep 23, 2026, Llama 3.2 3B Instruct is dominated for Balanced on LMArena. DeepSeek V4 Flash 0423 scores 322.4 higher and costs 14% less, with a covering envelope. 13 of 136 rated, priced standard models are undominated. [Inspect model evidence](/models/meta-llama__llama-3.2-3b-instruct/) [Compare differences & requirements](/compare/?models=meta-llama%2Fllama-3.2-3b-instruct%2Cdeepseek%2Fdeepseek-v4-flash) Current model Llama 3.2 3B Instruct Balanced 3 tokens in per 1 out Balanced Summarise Chat Code gen Agentic Monthly spend (USD) Used only to say how many months a named switching cost would take to earn back. Leave blank to skip. Switching cost (USD) DeepSeek V4 Flash 0423 is both better and cheaper than Llama 3.2 3B Instruct: 322.4 points higher on LMArena and 14% less per million tokens, $0.02 cheaper at this mix. [Llama 3.2 3B Instruct](https://undominated.ai/models/meta-llama__llama-3.2-3b-instruct/) LMArena Elo · higher is better Scale starts at 1040 Elo ** 1109.4 Effective $/M · Balanced · lower is better ** $0.120/M [DeepSeek V4 Flash 0423](https://undominated.ai/models/deepseek__deepseek-v4-flash/) LMArena Elo · higher is better Scale starts at 1040 Elo ** 1431.8 Effective $/M · Balanced · lower is better ** $0.103/M Llama 3.2 3B Instruct takes text, returns up to 117,964 tokens from a 131,072-token context, and is listed by 2 sellers. Compared against 136 rated, priced models on this lens: 3 models dominate it and give up nothing, 5 more dominate it but give something up. Llama 3.2 3B Instruct scores 1109.4 at $0.12 per million tokens for this mix. ## Envelope-safe replacements Each row scores at least as high, costs no more, and covers this model’s context, output, modalities, tools, and reasoning. A cheaper narrower model is not listed here. | Model | LMArena | Effective $/M | You save | | --- | --- | --- | --- | | [DeepSeek V4 Flash 0423](/models/deepseek__deepseek-v4-flash/) | 1431.8 +322.4 | $0.103/M | 14% | | [Granite 4.2 8B](/models/ibm-granite__granite-4.2-8b/) | 1316.8 +207.4 | $0.107/M | 10% | | [Llama 3.1 8B Instruct](/models/meta-llama__llama-3.1-8b-instruct/) | 1186.5 +77.1 | $0.058/M | 52% | ## Higher score, lower price, named losses Not a drop-in. The loss is why these are not a recommendation. | Model | LMArena | Effective $/M | You give up | | --- | --- | --- | --- | | [Qwen3.5-Flash](/models/qwen__qwen3.5-flash-02-23/) | 1397.7 | $0.114/M | max output | | [Qwen3 30B A3B Instruct 2507](/models/qwen__qwen3-30b-a3b-instruct-2507/) | 1383.6 | $0.084/M | max output | | [Gemma 3 12B](/models/google__gemma-3-12b-it/) | 1334.2 | $0.075/M | max output | | [Gemma 3 4B](/models/google__gemma-3-4b-it/) | 1290.7 | $0.063/M | max output | | [gpt-oss-20b](/models/openai__gpt-oss-20b/) | 1287.3 | $0.036/M | max output | 136 rated, priced standard models. Chartreuse is the frontier. Cobalt is the model you named, when it is not on the staircase. Copy watch URL [Open the frontier](/frontier/) The link keeps the model and mix, using the current catalogue. Monthly spend and switching cost stay in this browser tab and are left out of the link. ## Saved decision references Save the model, workload, catalogue date and capability-preservation rule in this browser. Spend, switching cost and bill contents are not saved. No account or notifications. Constraint: replacements must preserve the model’s capabilities; any losses remain named trade-offs. Historical decisions cannot be fully replayed from saved references: past prices, scores and capability evidence are not stored. Save this reference ## Continue your investigation - [Build a shortlist](/compare/) - [Inspect alternatives](/alternatives/) - [Check billing conditions](/traps/) - [Read model evidence](/models/)