Skip to main content
BenchLM
Data
Recommendation
Ranking data as of

Best Factuality AI Models in 2026

As of November 19, 2026, the top model among factuality AI models on the BenchLM leaderboard is Claude Fable 5.2 with a score of 73.2.

Bottom line: Factuality benchmarks are intentionally narrow — SimpleQA and HLE-no-tools are the primary signals. The category is still maturing, so treat the ranking as a shortlist.

Full Rankings (70 ranked)

1
Claude Fable 5.2
Anthropic·Proprietary·1M

73.2

sourced avg

vs #2
2
Claude Opus 5.5
Anthropic·Proprietary·1M

64.4

sourced avg

vs #3
3
GPT-4.5
OpenAI·Proprietary·128K

62.5

sourced avg

vs #4
4
Claude Fable 5.1
Anthropic·Proprietary·1M

60.9

sourced avg

vs #5
5
Claude Mythos 5
Anthropic·Proprietary·1M+

59

sourced avg

vs #6
6
Claude Sonnet 5.5
Anthropic·Proprietary·1M

56.9

sourced avg

vs #7
7
Claude Opus 5
Anthropic·Proprietary·

56.3

sourced avg

vs #8
8
Qwen3 235B 2507
Alibaba·Open Weight·128K

54.3

sourced avg

vs #9
9
Muse Spark 1.1
Meta·Proprietary·1M

52.2

sourced avg

vs #10
10
Qwen4 Max
Alibaba·Proprietary·1M

51.7

sourced avg

vs #11
11
Kimi K3.1 Pro
Moonshot AI·Open Weight·1.05M

51.1

sourced avg

vs #12
12
Sakana Fugu-Ultra
Sakana AI·Proprietary·1M

50

sourced avg

vs #13
13
Claude Opus 4.8
Anthropic·Proprietary·1M

49.8

sourced avg

vs #14
14
Pareto 26.9
Unbiased·Proprietary·N/A

49

sourced avg

vs #15
15
Sakana Fugu
Sakana AI·Proprietary·1M

47.2

sourced avg

vs #16
16
Claude Opus 4.7
Anthropic·Proprietary·1M

46.9

sourced avg

vs #17
17
ERNIE 5.5 Flash Preview
Baidu·Pending·1M

46.4

sourced avg

vs #18
18
Claude Haiku 5.5
Anthropic·Proprietary·1M

45.9

sourced avg

vs #19
19
Gemini 3.1 Pro
Google·Proprietary·1M

45.4

sourced avg

vs #20
20
Ornith-1.5-397B
Ornith AI·Open Weight·262K

44.6

sourced avg

vs #21
21
Inkling 2 Small
Thinking Machines Lab·Proprietary·1M

44.1

sourced avg

vs #22
22
Grok 3 [Beta]
xAI·Proprietary·128K

43.6

sourced avg

vs #23
23
Qwen3.8 Max
Alibaba·Open Weight·1M

43.6

sourced avg

vs #24
24
Kimi K3
Moonshot AI·Open Weight·1.05M

43.5

sourced avg

vs #25
25
Hy4 preview
Tencent·Open Weight·1M

43.4

sourced avg

vs #26
26
MiniMax M3.1 Flash
MiniMax·Proprietary·1M

43.4

sourced avg

vs #27
27
Claude Sonnet 5
Anthropic·Proprietary·1M

43.2

sourced avg

vs #28
28
GPT-5.5 Pro
OpenAI·Proprietary·1M

43.1

sourced avg

vs #29
29
Muse Spark
Meta·Proprietary·262K

42.8

sourced avg

vs #30
30
GPT-5.4 Pro
OpenAI·Proprietary·1.05M

42.7

sourced avg

vs #31
31
GPT-5.5
OpenAI·Proprietary·1M

41.4

sourced avg

vs #32
32
GLM-5.2
Z.AI·Open Weight·1M

40.5

sourced avg

vs #33
33
Claude Opus 4.6
Anthropic·Proprietary·1M

40

sourced avg

vs #34
34
GPT-5.4
OpenAI·Proprietary·1.05M

39.8

sourced avg

vs #35
35
Beam 1.0 501B-A23B
Reflection AI·Open Weight·1M

36

sourced avg

vs #36
36
Beam 1.0 501B-A23B FP8
Reflection AI·Open Weight·1M

36

sourced avg

vs #37
37
Qwen3.8-Flash-Next
Alibaba·Open Weight·262K

35.9

sourced avg

vs #38
38
Beam 1.0 501B-A23B NVFP4
Reflection AI·Open Weight·1M

34.8

sourced avg

vs #39
39
MiMo-V2.5-Pro
Xiaomi·Proprietary·1M

34

sourced avg

vs #40
40
Nemotron 3.5 Super VL 120B A12B
NVIDIA·Open Weight·1M

31.8

sourced avg

vs #41
41
Grok 4.20
xAI·Proprietary·2M

31.6

sourced avg

vs #42
42
Inkling-Small
Thinking Machines Lab·Open Weight·1M

31.6

sourced avg

vs #43
43

31.4

sourced avg

vs #44
44
Kimi K2
Moonshot AI·Proprietary·128K

31

sourced avg

vs #45
45
Qwen3.8-27B
Alibaba·Open Weight·262K

30.8

sourced avg

vs #46
46
DeepSeek-R1
DeepSeek·Open Weight·128K

30.1

sourced avg

vs #47
47
Inkling
Thinking Machines Lab·Open Weight·1M

30

sourced avg

vs #48
48
Gemini 2.0 Flash
Google·Proprietary·1M

29.9

sourced avg

vs #49
49
Solar Open 2
Upstage·Open Weight·1M

28.8

sourced avg

vs #50
50
Nemotron 3.5 Bolt 74B-A9B BF16
NVIDIA·Open Weight·256K

28.7

sourced avg

vs #51
51
GPT-5.4 mini
OpenAI·Proprietary·400K

28.2

sourced avg

vs #52
52
DeepSeek-R1-0528
DeepSeek·Open Weight·128K

27.8

sourced avg

vs #53
53
Nemotron 3 Ultra
NVIDIA·Open Weight·1M

26.7

sourced avg

vs #54
54
Ornith-1.5-35B-A3B
Ornith AI·Open Weight·262K

25.6

sourced avg

vs #55
55
Nemotron 3.5 Bolt 74B-A9B
NVIDIA·Open Weight·256K

25.3

sourced avg

vs #56
56
DeepSeek V3
DeepSeek·Open Weight·128K

24.9

sourced avg

vs #57
57
GPT-5.4 nano
OpenAI·Proprietary·400K

24.3

sourced avg

vs #58
58
MiniMax-Text-01
MiniMax·Open Weight·1M

23.7

sourced avg

vs #59
59
o4-mini
OpenAI·Proprietary·200K

20.2

sourced avg

vs #60
60
Ornith-1.5-9B
Ornith AI·Open Weight·262K

20.2

sourced avg

vs #61
61
Gemma 4 31B
Google·Open Weight·256K

19.5

sourced avg

vs #62
62
Spark-X2.6-4B
SparkLLM·Open Weight·1M

17.9

sourced avg

vs #63
63
Gemini 2.0 Flash-Lite
Google·Proprietary·1M

16.5

sourced avg

vs #64
64
Gemini 2.5 Flash-Lite
Google·Proprietary·1M

10.7

sourced avg

vs #65
65

10.5

sourced avg

vs #66
66
Mistral Small 3.1
Mistral·Open Weight·128K

10.4

sourced avg

vs #67
67
Gemma 4 26B A4B
Google·Open Weight·256K

8.7

sourced avg

vs #68
68
Gemini 1.5 Flash
Google·Proprietary·1M

8.6

sourced avg

vs #69
69
o1-mini
OpenAI·Proprietary·128K

7.6

sourced avg

vs #70
70
Gemma 4 12B
Google·Open Weight·256K

5.2

sourced avg

How to choose

Key Takeaways

The top model on this sourced reporting-family slice is Claude Fable 5.2 by Anthropic with an average of 73.2.

The best open-weight model is Qwen3 235B 2507 at position #8.

70 models are listed with sourced benchmark coverage in this reporting family.

Score in Context

What these scores mean

This is a reporting family ranking, not a weighted category. It averages sourced factuality benchmarks to give a focused view of this capability.

Known limitations

Models must have sourced results on at least a quarter of the benchmarks in this family to be included. Coverage varies — a model with 2 benchmark scores is less reliable than one with 5.

About this ranking

Ranking data as of November 19, 2026

This reporting page is intentionally narrow. It focuses on currently tracked sourced factuality signals such as SimpleQA, HLE without tools, and multimodal factuality. It is a reporting page, not a mature weighted category.

This page ranks models using only sourced factuality benchmarks in the reporting family.

Claude Fable 5.2 leads this ranking with a score of 73.2, followed by Claude Opus 5.5 (64.4) and GPT-4.5 (62.5). There is a significant gap between the leading models and the rest of the field.

The best open-weight option is Qwen3 235B 2507 (ranked #8 with a score of 54.3). While proprietary models lead, open-weight options are within striking distance for teams willing to trade a few points of performance for full model control.

This ranking uses provisional overall weighted scores from the active scoring formula. For detailed model profiles, click any model name above. To compare two specific models head-to-head, use the "vs #" links.

Know when it’s worth switching models

The model to choose, the cheaper alternative, and the release we would wait on.

Read a sample issue

Join 7,000+ subscribers.

One email each week. Unsubscribe anytime.