The newest Grok model is Grok 4.7, released on September 21, 2026. The line's usual gap is about 59 days and 33 have passed, so it is mid-cycle. xAI has officially announced Grok 4.8; on Sep 14, 2026 it said: “Grok 4.8, which is a 2.5T model trained with our new C++ software stack, will finish training this week and start RL”.
How long it has been since the last release against the line's usual gap and its recent ones — the rhythm, not a date.
Latest: Grok 4.7 · Sep 21, 2026
“Grok 4.8, which is a 2.5T model trained with our new C++ software stack, will finish training this week and start RL” — xAI, Sep 14, 2026 · Elon Musk on X
What changed in Grok 4.7 →Every Monday, only when something shipped: the AI models that came out, what each one changed, and which lines are due next.
We email you a link to confirm first · unsubscribe anytime · read past issues · about the newsletter · RSS for the Grok line
Newest first — what each version changed versus the one before, in xAI's own numbers.
A larger base model than Grok 4.6, trained longer on multi-hour agentic tasks, at the same $2 / $6 price and the same 500k-token context.
Compared with Grok 4.6 · the model it replaces
Solves 7 in 10 real-world software tasks
DeepSWE v1.1 — fixing real bugs in open-source code
71.0%
▲ up from 65.2% in xAI's Grok 4.7 table · +5.8 points · both at high effort
Scores 2 points higher on the all-round intelligence index
Artificial Analysis Intelligence Index v4.3.2 — one score averaging 10 hard tests
46
▲ up from 44 · +2 points
Costs the same as the model it replaces
API price per 1M tokens — input / output
$2 / $6
same as Grok 4.6 · higher rate above 200k prompt tokens
Builds on Grok 4.5 with a longer extra training run; adds an xhigh reasoning-effort level, and cached input now costs $0.50 per million tokens, up from $0.30.
Compared with Grok 4.5 · the model it replaces
Solves about 2 in 3 real-world software tasks
DeepSWE v1.1 — fixing real bugs in open-source code
65.9%
▲ up from 54.0% · +11.9 points
Scores 5 points higher on the all-round intelligence index
Artificial Analysis Intelligence Index v4.3.2 — one score averaging 10 hard tests
44
▲ up from 39 · +5 points
Costs the same as the model it replaces
API price per 1M tokens — input / output
$2 / $6
same as Grok 4.5
Added configurable reasoning effort (low, medium, high) with a 500K-token context window, priced at $2 per million input tokens and $6 per million output tokens.
xAI's update to Grok 4.20: it can now hold a whole book of text at once, watch video, and costs 40% less for the text you send in — with a few points higher on a broad skills test.
Compared with Grok 4.20 · the model it replaces
Can hold a whole book in mind at once
Context window — how much text it can read in one go
1M tokens
✦ new in Grok 4.3
Can now watch video, not just read text and see images
Video input — understanding moving footage directly
native
✦ new in Grok 4.3
Costs 40% less for the text you send it
Input price — what developers pay for the text they send in
40% cheaper
▲ vs Grok 4.20 · input only
xAI's new flagship Grok: the first version that retrains itself every week from real-world feedback, and in March 2026 it held the record on Artificial Analysis's test for admitting it doesn't know instead of making up an answer.
Compared with Grok 4.1 · the version it replaces
Retrains itself every week from real-world feedback
Update cadence — how often the model is refreshed
weekly
✦ new in Grok 4.20 · first Grok to do this
When it didn't know an answer, it said so 78% of the time instead of making one up
Artificial Analysis Omniscience test — how often it admits not knowing
78%
✦ a record when reported in March 2026 · other models have beaten it since
Scores higher than Grok 4.1 when people compare answers blind
Arena Text leaderboard — people pick the better answer without knowing the model
1475
▲ up from 1465 for Grok 4.1 Thinking · +10 Elo · #34 on Arena, Sep 2026
xAI scaled up reinforcement-learning training for Grok 4 by over an order of magnitude in compute, with native tool use and real-time search and a 256,000-token API context window, at the same $3 / $15 price as Grok 3.
Compared with Grok 3 · the model it replaces
Solves about 1 in 6 abstract reasoning puzzles
ARC-AGI-2 — visual puzzles that test abstract reasoning
15.9%
✦ nearly double Claude Opus 4's 8.6% in xAI's chart
Costs the same as Grok 3
API price per 1M tokens — input / output
$3 / $15
same as Grok 3