[good]

Best Value LLMs

Value score = quality per dollar. A model with a 7.5 quality score at $0.11/1M ranks higher than a model with an 8.0 quality score at $5.00/1M. Models with a genuinely free consumer product receive a boost — free is hard to beat on value.

1
Gemini 3 ProGooglefree consumer product
Quality: 8.8Free to use via web — no API required

Top overall value: highest quality-per-dollar ratio in this comparison.

10.0
value
Consumer plan: Google One AI Premium at $20/month
2
Llama 4 ScoutMetaopen weights
Quality: 7.5API cost: $0.11/1M blended via Self-hosted
10.0
value
3
Gemini 3 FlashGooglefree consumer product
Quality: 7.3Free to use via web — no API required
8.8
value
Consumer plan: Google One AI Premium at $20/month
4
DeepSeek V3.2DeepSeekfree consumer productopen weights
Quality: 6.3Free to use via web — no API required
7.6
value
5
GPT-5 miniOpenAI
Quality: 7.3API cost: $0.69/1M blended via OpenAI
3.2
value
Consumer plan: ChatGPT Free at $0/month
6
Llama 4 MaverickMetaopen weights
Quality: 4.4API cost: $0.44/1M blended via Self-hosted
2.6
value
7
Mistral Large 3Mistralopen weights
Quality: 4.6API cost: $0.75/1M blended via Self-hosted
1.9
value
Consumer plan: Le Chat Pro at $15/month
8
GPT-5.2OpenAI
Quality: 8.3API cost: $4.81/1M blended via OpenAI
1.2
value
Consumer plan: ChatGPT Plus at $20/month
9
Claude Sonnet 4.6Anthropic
Quality: 8.0API cost: $6.00/1M blended via Anthropic
1.0
value
Consumer plan: Claude Pro at $20/month
10
Grok 4.1xAI
Quality: 8.0API cost: $6.00/1M blended via xAI
1.0
value
Consumer plan: X Premium at $8/month
11
Claude Opus 4.6Anthropic
Quality: 7.5API cost: $10.00/1M blended via Anthropic
0.8
value
Consumer plan: Claude Pro (limited Opus access) at $20/month

How the value score is calculated

Value score = quality rating (0–10) divided by the log of blended API cost. Log scale means cost differences matter more at the low end: going from $0.11 to $0.50 hurts more than going from $4.50 to $5.00. Models with a genuinely free consumer product (Gemini, DeepSeek web) receive a 20% quality bonus, capped at 10 — because for everyday users who will never pay per token, free is exceptional value regardless of API cost.

Value scores are not comparable across sessions — they depend on the current field. Adding a cheaper or more capable model will shift rankings. See how we rate for the full quality scoring methodology.

Free web chatFree API tierBudget API (under $1/1M)