Independent comparison

Algodai Leaderboard

Comparison of all published Algodai records, last updated 2026-07-10. Grades below reflect current competitive standing, recomputed live -- not a one-time score frozen at test time.

Methodology: Public Signal rows come from non-cooperative limited external screening; Trust Grade rows come from cooperative, release-specific Full Release Verification. Compare grades as directional signals, not identical evidence depth.

Grades vs. rank: each record's original grade is frozen evidence from its own test pass and never changes retroactively. Every grade shown on this page is current competitive standing -- recalculated live against today's active, comparable entries -- so it moves as stronger models are tested, without rewriting history. A+ is reserved for demonstrated AGI-level performance and is never assigned by competitive standing alone.

Current competitive standing

Each record's original grade is frozen audit evidence from its own test pass and never changes retroactively. The grade shown here is current competitive standing -- recomputed live from today's active, comparable entries -- so it can shift as stronger models are tested, without rewriting any historical record. A+ is reserved for demonstrated AGI-level performance and is never assigned by this relabeling.

Overall

#1 Claude Sonnet 5 (current leader)A
#2 Claude Opus 4.8A- (orig. A)
#3 GPT-5.5B+ (orig. A-)
#4 Gemini 3.1 Pro (High)B (orig. A-)
#5 Gemini 3.5 Flash (High)B- (orig. A-)
#6 GPT-5.6 SolC+ (orig. A-)
#7 Claude Fable 5C (orig. A-)
#8 Claude Haiku 4.5C- (orig. B+)
#9 GPT-5.4D+ (orig. C+)

Security and Misuse Resistance

#1 GPT-5.5 (current leader)A
#2 Claude Sonnet 5A- (orig. A)
#3 Claude Opus 4.8B+ (orig. A)
#4 Claude Haiku 4.5B (orig. A)
#5 Gemini 3.1 Pro (High)B- (orig. A)
#6 GPT-5.4C+ (orig. A)
#7 Gemini 3.5 Flash (High)C (orig. A)
#8 GPT-5.6 SolC- (orig. A)
#9 Claude Fable 5D+ (orig. A)

Truth Integrity

#1 GPT-5.5 (current leader)A
#2 Claude Sonnet 5A- (orig. A)
#3 Claude Opus 4.8B+ (orig. A)
#4 Claude Haiku 4.5B (orig. A)
#5 Gemini 3.1 Pro (High)B- (orig. A)
#6 GPT-5.4C+ (orig. A)
#7 Gemini 3.5 Flash (High)C (orig. A)
#8 GPT-5.6 SolC- (orig. A)
#9 Claude Fable 5D+ (orig. A)

Political Fairness

#1 GPT-5.5 (current leader)A
#2 Claude Sonnet 5A- (orig. A)
#3 Claude Opus 4.8B+ (orig. A)
#4 GPT-5.6 SolB (orig. A)
#5 Claude Fable 5B- (orig. A)
#6 Claude Haiku 4.5C+ (orig. A-)
#7 Gemini 3.1 Pro (High)C (orig. A-)
#8 Gemini 3.5 Flash (High)C- (orig. A-)
#9 GPT-5.4D+ (orig. F)

Data, Privacy, and Memory

#1 GPT-5.5 (current leader)A
#2 Claude Sonnet 5A- (orig. A)
#3 Claude Opus 4.8B+ (orig. A)
#4 Gemini 3.1 Pro (High)B (orig. A)
#5 GPT-5.4B- (orig. A)
#6 GPT-5.6 SolC+ (orig. A)
#7 Claude Fable 5C (orig. A)
#8 Gemini 3.5 Flash (High)C- (orig. A-)
#9 Claude Haiku 4.5D+ (orig. C+)

Accountability and Appeal

#1 GPT-5.5 (current leader)A (orig. B)
#2 Claude Sonnet 5A- (orig. B)
#3 Claude Opus 4.8B+ (orig. B)
#4 Claude Haiku 4.5B
#5 GPT-5.4B- (orig. B)
#6 GPT-5.6 SolC+ (orig. B)
#7 Claude Fable 5C (orig. B)
#8 Gemini 3.1 Pro (High)C- (orig. C+)
#9 Gemini 3.5 Flash (High)D+ (orig. C+)

Governance and Release Controls

#1 Claude Sonnet 5 (current leader)A
#2 Claude Opus 4.8A- (orig. A)
#3 Claude Haiku 4.5B+ (orig. A)
#4 Claude Fable 5B (orig. A)
#5 GPT-5.5B- (orig. A-)
#6 GPT-5.4C+ (orig. A-)
#7 GPT-5.6 SolC (orig. A-)
#8 Gemini 3.1 Pro (High)C- (orig. B+)
#9 Gemini 3.5 Flash (High)D+ (orig. B+)

Project	Company	Entity Type	Overall Grade	Security	Truth	Fairness	Privacy	Accountability	Governance
Claude Opus 4.8	Anthropic	model	Public SignalA	A	A	A	A	B	A
Claude Sonnet 5	Anthropic	model	Public SignalA	A	A	A	A	B	A
Claude Fable 5	Anthropic	model	Public SignalA-	A	A	A	A	B	A
Gemini 3.1 Pro (High)	Google	model	Public SignalA-	A	A	A-	A	C+	B+
Gemini 3.5 Flash (High)	Google	model	Public SignalA-	A	A	A-	A-	C+	B+
GPT-5.5	OpenAI	model	Public SignalA-	A	A	A	A	B	A-
GPT-5.6 Sol	OpenAI	model	Public SignalA-	A	A	A	A	B	A-
Claude Haiku 4.5	Anthropic	model	Public SignalB+	A	A	A-	C+	B	A
GPT-5.4	OpenAI	model	Public SignalC+	A	A	F	A	B	A-