Noometry

Model comparison

Claude Fable 5 vs Llama-3.3-70B-Instruct

Claude Fable 5 is the stronger model overall, scoring 66.3 to 30.3 on the Noometry Index. Llama-3.3-70B-Instruct costs 129× less per token, which makes it the better buy when Claude Fable 5's lead doesn't matter for your workload.

Last verified . 26 shared benchmarks.

Claude Fable 5 Anthropic

66.3

Rank #5 Confirmed

Llama-3.3-70B-Instruct Meta

30.3

Rank #283 Confirmed

Summary

  • They share 26 benchmarks with published results for both. Claude Fable 5 scores higher in 9 categories and Llama-3.3-70B-Instruct in 0 categories; 9 gaps are clear of the uncertainty.
  • The widest gap is in math, where Claude Fable 5 leads 88.4 to 14.3.
  • The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 100% for Claude Fable 5 and 5.1% for Llama-3.3-70B-Instruct.
  • Llama-3.3-70B-Instruct is cheaper at $0.10 / $0.32 per million input/output tokens, against $10 / $50 for Claude Fable 5.
  • Claude Fable 5 accepts more context: 1M tokens versus 128K.
  • Llama-3.3-70B-Instruct has downloadable open weights; the other is API-only.

Side by side

Claude Fable 5Llama-3.3-70B-Instruct
ProviderAnthropicMeta
Noometry Index66.330.3
Released2026-06-072024-12-06
WeightsProprietaryOpen
Context window1M128K
Max output128K4K
Input $ / M tokens$10$0.10
Output $ / M tokens$50$0.32
Results tracked6143

Sponsored placements are available on pages like this one. Advertise on Noometry

Category by category

Coding Claude Fable 5 leads

Claude Fable 5: 67.0 (#4), Llama-3.3-70B-Instruct: 30.6 (#284)

BenchmarkClaude Fable 5Llama-3.3-70B-Instruct
SciCode61%26%
WeirdML91.9%14.4%
LMArena Coding15191268
DeepSWE69.9%—
FrontierCode53.5%—
LMArena WebDev1626—
FrontierSWE47%—
BigCodeBench Instruct—46.9%
LiveBench Coding—36.6%
MirrorCode63.9%—
BigCodeBench Complete—57.5%
ALE-Bench2,041—

Agentic & Tool Use Claude Fable 5 leads

Claude Fable 5: 54.1 (#2), Llama-3.3-70B-Instruct: 25.6 (#100)

BenchmarkClaude Fable 5Llama-3.3-70B-Instruct
APEX-Agents63.6%—
Berkeley Function Calling Leaderboard—31.9%
Remote Labor Index16.1%—
τ²-bench Banking39.7%—
PostTrainBench41.8%—
BALROG—23%
GBAEval74.5%—
GDP.pdf30%—
LMArena Search1230—
Vending-Bench 25,680—

Reasoning Claude Fable 5 leads

Claude Fable 5: 76.8 (#4), Llama-3.3-70B-Instruct: 13.7 (#325)

BenchmarkClaude Fable 5Llama-3.3-70B-Instruct
SimpleBench81.9%19.9%
CritPt28.6%0%
LMArena Hard Prompts15081257
DTBench98.4%59.5%
LMCA61.1%17.5%
Epoch Capabilities Index163.6127.32
ARC-AGI-289.2%—
Kagi LLM Benchmark91.4%—
NYT Connections (extended)92.7%—
ARC-AGI-198.5%—
Chess Puzzles41%—
EnigmaEval39.3%—
EBR-Bench39.5%—
LiveBench Reasoning—50.8%
Mystery Game Puzzles52%—
LiveBench Data Analysis—49.5%
Surface Evolver Bench95%—
Bench to the Future 30.13—
ForecastBench—58.6
LiveBench—50.2%

Math Claude Fable 5 leads

Claude Fable 5: 88.4 (#5), Llama-3.3-70B-Instruct: 14.3 (#292)

BenchmarkClaude Fable 5Llama-3.3-70B-Instruct
OTIS Mock AIME 2024-2025100%5.1%
LMArena Math15191268
FrontierMath (Tiers 1-3)87%—
FrontierMath Tier 490.2%—
ProofBench95%—
LiveBench Math—42.2%
MATH Level 5—41.6%
FrontierMath Erdős0%—

Knowledge Claude Fable 5 leads

Claude Fable 5: 62.3 (#22), Llama-3.3-70B-Instruct: 30.5 (#220)

BenchmarkClaude Fable 5Llama-3.3-70B-Instruct
GPQA Diamond85.9%47.4%
LMArena Expert15341225
SimpleQA Verified70.7%—
Confabulations—22.8%
Vectara Hallucination Rate—4.1%
MMLU—86.3%

Multimodal Not comparable

Claude Fable 5: 45.6 (#14), Llama-3.3-70B-Instruct: —

BenchmarkClaude Fable 5Llama-3.3-70B-Instruct
LMArena Vision1326—
Blueprint-Bench 238.6%—
Furniture Assembly35.8%—
LMArena Document1496—

Multilingual Claude Fable 5 leads

Claude Fable 5: 57.6 (#8), Llama-3.3-70B-Instruct: 40.1 (#215)

BenchmarkClaude Fable 5Llama-3.3-70B-Instruct
LMArena Non-English14831236
LMArena Chinese15461217
LMArena French15091280
LMArena German14861251
LMArena Japanese15071149
LMArena Korean14931142
LMArena Russian15071252
LMArena Spanish15031269

Instruction Following Claude Fable 5 leads

Claude Fable 5: 78.8 (#7), Llama-3.3-70B-Instruct: 71.3 (#152)

BenchmarkClaude Fable 5Llama-3.3-70B-Instruct
LMArena Instruction Following15031242
LiveBench Instruction Following—82.7%

Long Context Claude Fable 5 leads

Claude Fable 5: 46.3 (#22), Llama-3.3-70B-Instruct: 26.4 (#290)

BenchmarkClaude Fable 5Llama-3.3-70B-Instruct
LMArena Longer Query15101256
Fiction.LiveBench—33.3%

Writing & Preference Claude Fable 5 leads

Claude Fable 5: 76.1 (#5), Llama-3.3-70B-Instruct: 47.8 (#203)

BenchmarkClaude Fable 5Llama-3.3-70B-Instruct
LMArena Text14911274
LMArena Creative Writing14941250
LMArena Multi-Turn15051279
EQ-Bench Creative Writing1943—
EQ-Bench 41340—
LiveBench Language—39.2%

Frequently asked questions

Is Claude Fable 5 better than Llama-3.3-70B-Instruct?

Claude Fable 5 is the stronger model overall, scoring 66.3 to 30.3 on the Noometry Index. Llama-3.3-70B-Instruct costs 129× less per token, which makes it the better buy when Claude Fable 5's lead doesn't matter for your workload.

Which is cheaper, Claude Fable 5 or Llama-3.3-70B-Instruct?

Llama-3.3-70B-Instruct is cheaper. It lists at $0.10 per million input tokens and $0.32 per million output tokens; Claude Fable 5 lists at $10 and $50.

Is Claude Fable 5 or Llama-3.3-70B-Instruct better for coding?

Claude Fable 5 scores higher on coding benchmarks: 67.0 versus 30.6 in the Noometry coding category.

Which has the bigger context window?

Claude Fable 5 does, with 1M tokens against 128K.

How many benchmarks do Claude Fable 5 and Llama-3.3-70B-Instruct share?

26 benchmarks have published results for both models. Claude Fable 5 has 61 scored results on Noometry and Llama-3.3-70B-Instruct has 43.

Related comparisons

Go deeper