Skip to main content
FreeAIList

Find a model for your task

Choose your task and access needs to find models with documented options.

Code generation, debugging, refactoring, tool calling, and API integration.

Documented recurring free API plan with no mandatory subscription.

Balanced catalog order based on verified publication evidence.

How matching works

Recommendations use published access facts and rule-based matching. Benchmark observations and caveats are shown separately. We do not generate ranking scores. Check each model’s linked sources before choosing a provider.

Compare modelsMethodology

Matching models 4 found

Compare first 3 matches

Google

Gemini 3.8 Flash

A multimodal Flash model with a 1,048,576-token input window and thinking support.

Version
gemini-3.8-flash
Capabilities
text, image input, audio/video context, tool use, thinking reasoning

Documented access

  • via Google AI Studio & Gemini API: Free input and output tokens; check AI Studio for current project and model quotas. (1,048,576 input / 65,536 output tokens context; Reset period unknown; No payment card required; Source verified).

    Gemini API pricing

Last verified

Important limits

  • Google lists a Free Tier; exact rate limits vary by model and project
  • Daily request quotas reset at midnight Pacific time
Benchmark observations (1)

  • OTIS Mock AIME 2024–2025

    99%

    accuracy · Independent evaluation

    Source verified

    Epoch AI administered the 45-question OTIS Mock AIME 2024–2025 math evaluation. The linked model scorecard reports this accuracy. See https://epoch.ai/benchmarks/otis-mock-aime-2024-2025 for test setup and https://epoch.ai/benchmarks/use-this-data for data license and citation.

    Epoch AI model scorecard (CC BY 4.0)CC BY 4.0

Google

Gemini 3.7 Flash

A multimodal Flash model with a 1,048,576-token input window and thinking support.

Version
gemini-3.7-flash
Capabilities
text, image input, audio/video context, tool use, thinking reasoning

Documented access

  • via Google AI Studio & Gemini API: Free input and output tokens; check AI Studio for current project and model quotas. (1,048,576 input / 65,536 output tokens context; Reset period unknown; No payment card required; Source verified).

    Gemini API pricing

Last verified

Important limits

  • Google lists a Free Tier; exact rate limits vary by model and project
  • Daily request quotas reset at midnight Pacific time
Benchmark observations (1)

  • OTIS Mock AIME 2024–2025

    97%

    accuracy · Independent evaluation

    Source verified

    Epoch AI administered the 45-question OTIS Mock AIME 2024–2025 math evaluation. The linked model scorecard reports this accuracy. See https://epoch.ai/benchmarks/otis-mock-aime-2024-2025 for test setup and https://epoch.ai/benchmarks/use-this-data for data license and citation.

    Epoch AI model scorecard (CC BY 4.0)CC BY 4.0

Google

Gemini 3.6 Flash

A multimodal Flash model with a 1,048,576-token input window and thinking support.

Version
gemini-3.6-flash
Capabilities
text, image input, audio/video context, tool use, thinking reasoning

Documented access

  • via Google AI Studio & Gemini API: Free input and output tokens; check AI Studio for current project and model quotas. (1,048,576 input / 65,536 output tokens context; Reset period unknown; No payment card required; Source verified).

    Gemini API pricing

Last verified

Important limits

  • Google lists a Free Tier; exact rate limits vary by model and project
  • Daily request quotas reset at midnight Pacific time
Benchmark observations (1)

  • OTIS Mock AIME 2024–2025

    94%

    accuracy · Independent evaluation

    Source verified

    Epoch AI administered the 45-question OTIS Mock AIME 2024–2025 math evaluation. The linked model scorecard reports this accuracy. See https://epoch.ai/benchmarks/otis-mock-aime-2024-2025 for test setup and https://epoch.ai/benchmarks/use-this-data for data license and citation.

    Epoch AI model scorecard (CC BY 4.0)CC BY 4.0

OpenAI

GPT OSS 20B

A smaller open-weight reasoning model documented in Groq’s hosted model catalog.

Version
openai/gpt-oss-20b
Capabilities
text, reasoning, coding

Documented access

  • via GroqCloud API: Groq documents a developer plan and model-specific limits; exact free allowance is account-specific. (131,072 tokens context; daily reset; No payment card required; Source verified).

    GroqCloud models

Last verified

Important limits

  • The provider catalog is the evidence for access, not a universal free quota
  • Model availability and limits can change
Benchmark observations (4)

  • GPQA

    65.4%

    accuracy · Independent evaluation

    Source verified

    Swallow LLM Leaderboard v2, swallow-evaluation-instruct v202604. Post-trained English results on a 0–1 scale, converted to percent here. See the linked result and https://swallow-llm.github.io/leaderboard/about.en.html for methods and CC BY 4.0 terms.

    Swallow LLM Leaderboard (CC BY 4.0)CC BY 4.0

  • LiveCodeBench

    59.9%

    pass@1 · Independent evaluation

    Source verified

    Swallow LLM Leaderboard v2, swallow-evaluation-instruct v202604. Post-trained English results on a 0–1 scale, converted to percent here. See the linked result and https://swallow-llm.github.io/leaderboard/about.en.html for methods and CC BY 4.0 terms.

    Swallow LLM Leaderboard (CC BY 4.0)CC BY 4.0

  • MMLU Pro

    74.8%

    accuracy · Independent evaluation

    Source verified

    Swallow LLM Leaderboard v2, swallow-evaluation-instruct v202604. Post-trained English results on a 0–1 scale, converted to percent here. See the linked result and https://swallow-llm.github.io/leaderboard/about.en.html for methods and CC BY 4.0 terms.

    Swallow LLM Leaderboard (CC BY 4.0)CC BY 4.0

  • OTIS Mock AIME 2024–2025

    65%

    accuracy · Independent evaluation

    Source verified

    Epoch AI administered the 45-question OTIS Mock AIME 2024–2025 math evaluation. The linked model scorecard reports this accuracy. See https://epoch.ai/benchmarks/otis-mock-aime-2024-2025 for test setup and https://epoch.ai/benchmarks/use-this-data for data license and citation.

    Epoch AI model scorecard (CC BY 4.0)CC BY 4.0

Transparency standard

Why FreeAIList does not manufacture a token-efficiency ranking

Visitors often ask: “Which model uses fewer tokens while still getting the job done?”

FreeAIList cannot determine which model uses fewer tokens because comparable, cross-provider benchmarks for prompt expansion, tokenizer density, and task quality do not exist. Tokenizers vary widely across families (a prompt that is 100 tokens in one model may be 140 in another), and output length depends on system prompts and thinking modes.

Rather than inventing an unverified efficiency number, we advise running a small test with 3 to 5 real prompts from your own workload to measure actual token counts and latency before committing to a provider.

Before you choose

Check Your Actual Free Limits

API Tiers

Free access can differ by model, provider, project, and region. Open the source linked on each recommendation and check the limit shown in your own account before depending on it.

  • • Google shows current project limits in AI Studio.
  • • Hosted providers may change which models qualify for starter usage.
  • • Unknown or disputed allowances are marked for rechecking.
FreeAIList Standard

How We Verify Models

  • ✓Rule-based matching: Recommendations use the published catalog and show their access evidence.
  • ✓Official documentation: Verified against provider pricing & rate-limit pages.
  • ✓Benchmark evidence: Numeric scores appear only when the cited source supports the result.
Read benchmark methodologyCompare all models