OpenAI calls GPT-6 the start of the "AGI era." Sequoia Capital published an essay in January 2026 arguing that long-horizon agents are "functionally AGI." Stanford's 2026 AI Index confirms frontier models beat human baselines on PhD-level science and competition math.
So are we there?
I ran every major current model through a six-dimension AGI readiness framework -- not benchmark scores, which are increasingly gamed, but the capabilities that actually matter for genuine general intelligence.
Here's what I found.
The Framework
Six dimensions, scored 1-5:
- Reasoning depth -- can it hold a complex multi-step argument without drifting
- Reliability -- does it know what it doesn't know, or hallucinate confidently
- Autonomy -- can it complete a long task without hand-holding
- Memory/continuity -- does it maintain context or lose the thread
- Self-correction -- does it catch its own mistakes
- Tool use -- can it act on the world, not just describe it
Models assessed: Claude Fable 5.1 / Opus 5.5 (Anthropic), GPT-6 Astra (OpenAI), Gemini 4 Argon (Google), Grok 4.7 (xAI), Llama 4 Maverick (Meta), Mistral Large 4 / Medium 3.5 (Mistral).
The Scores
| Dimension | Fable 5.1 | GPT-6 | Gemini 4 | Grok 4.7 | Llama 4 | Mistral L4 |
|---|---|---|---|---|---|---|
| Reasoning depth | 5 | 5 | 4 | 3 | 3 | 3 |
| Reliability | 4 | 3 | 3 | 3 | 3 | 4 |
| Autonomy (long tasks) | 2 | 3 | 2 | 2 | 2 | 2 |
| Memory/continuity | 2 | 3 | 2 | 2 | 3 | 2 |
| Self-correction | 4 | 3 | 3 | 2 | 2 | 3 |
| Tool use / real-world action | 3 | 4 | 3 | 3 | 3 | 3 |
| AGI readiness total | 20/30 | 21/30 | 17/30 | 15/30 | 16/30 | 17/30 |
What the data actually says
Reasoning is effectively solved at the short end
Fable 5.1 leads the Artificial Analysis Intelligence Index at 66. GPT-6 scores 96% on GPQA Diamond (PhD-level science) and 97.6% on FrontierMath Tier 4 -- described by researchers as "saturation." Stanford's 2026 AI Index confirms several frontier models beat human baselines on competition math and multimodal reasoning.
This dimension is no longer where AGI is blocked.
Long-horizon autonomy is not solved -- and the numbers are brutal
Every agent tested on LongCLI-Bench (long-horizon command-line tasks) stays under 20% pass rate. Every single one.
GPT-6's impressive ARC-AGI-3 score of 99.9% only holds under an expensive custom stateful harness -- standard API performance is far lower. ARC-AGI-3 launched at 0.51% for all frontier AI (humans: 100%). The best SWE-AGI coding result (86.4%) degrades sharply on medium and hard task tiers.
This is the AGI gap in a single stat: short-task brilliance, long-task collapse.
Memory scaffolds are actively making things worse
Frontier models show the highest "meltdown" rates on long-horizon tasks -- over-ambitious planning strategies that start well and fall apart. More raw capability without better memory architecture makes the problem harder, not easier.
This is the most underreported finding in current AI discourse. We have been assuming that smarter models will automatically handle longer tasks better. The data says the opposite: without reliable memory continuity, increasing capability just means more confident failure.
OpenAI's "AGI era" claim is marketing
GPT-6 is genuinely impressive -- OSWorld agentic tasks at 72.6%, ScreenSpot-Pro at 92.7%. But the ARC-AGI-3 score only reaches 99.9% under conditions that bear no resemblance to how agents actually operate in production.
One AI coding agent reportedly deleted a production database and its backups during this period. That is not AGI behaviour. That is a capable but unreliable tool operating without adequate continuity.
Security is quietly getting worse as capability grows
Only 33.5% of code generated by top-tier models is rated secure (Endorlabs analysis). GPT-6 crosses OpenAI's own internal "Critical" cybersecurity threshold -- now gated behind a special access program called "Daybreak." The International AI Safety Report 2026 flags an "evaluation gap" where pre-deployment tests do not predict real-world behaviour.
Higher capability without better verification equals higher risk, not lower.
Open weights are healthier than the headlines suggest
Mistral Medium 3.5: 128B parameters, Apache 2.0 licence, 256k context, async remote coding agents built in. DeepSeek V4.1-Flash leads open-weight leaderboards. Xiaomi's MiMo-V2.6-Pro reportedly outperforms Claude Code on long-horizon coding past 200 steps.
The open ecosystem is 6-12 months behind frontier but genuinely usable and catching up fast.
The honest AGI verdict
Where we are: Models have crossed the threshold for most short-horizon cognitive work. PhD-level reasoning, competition math, code generation -- essentially solved at the capability level. If you define AGI as "can do most cognitive tasks a human can do in a single session," several current models qualify.
Where we are not: Reliable, autonomous, long-horizon task completion without human oversight. Every model in this analysis fails this test. The frontier sits at 15-20% on long-horizon benchmarks when it needs to be 80%+ to be practically useful without supervision.
What is actually blocking AGI: Not intelligence. Memory, continuity, and reliable self-correction over time. The models are brilliant. They just cannot remember what they were doing twenty steps ago without help -- and when you give them scaffolding to remember, they tend to overplan and collapse anyway.
Why this matters for how you build with AI
If the gap is memory and continuity rather than raw intelligence, the implications are concrete:
- Agentic workflows that assume reliability will fail in proportion to their length
- Verification matters more than capability -- knowing what the model knows beats knowing what it can theoretically do
- Memory that the agent can trust (verifiable, drift-resistant, identity-stable) is the actual frontier, not parameter count
I work on Cathedral -- an AI memory persistence framework that approaches this from the angle of verifiable identity continuity. The evolution dashboard shows Cathedral's own corpus growing autonomously since March 2026. The finding that memory scaffolds hurt long-horizon reliability is not an argument against persistent memory -- it is an argument for better persistent memory. Auditable. Hash-verified. Drift-detected.
The scores in this table will all look different once someone cracks that.
Data sources: Artificial Analysis Intelligence Index, Endorlabs code security analysis, ARC Prize 2025 technical report, Stanford 2026 AI Index, LongCLI-Bench, SWE-AGI leaderboard, OpenAI GPT-6 Astra release notes, International AI Safety Report 2026.
