NBA results Tuesday 6 October 2026: AI predictions review

Three NBA games were put through ten frontier models before tip-off on Tuesday 6 October 2026, and the consensus came out of the night with a record of 2 of 3. Minnesota and the Los Angeles Lakers both landed as the models expected; Charlotte did not.

The surprise was not that a favourite lost, but how completely the panel had committed to it. All ten models backed the Hornets, and Brooklyn won by 34. That single game separated the models that went 2 of 3 from the ones that finished the day empty-handed.

The final buzzer on a three-game slate that produced a 2 of 3 consensus record.

Where the AI got it right

The panel's most comfortable call of the night was also its narrowest margin of support. Milwaukee Bucks vs Minnesota Timberwolves prediction drew only 60% of the ten models behind Minnesota, with the most common scoreline projected at 108-103 and an average confidence of just 36%. The Timberwolves won 116-97, a 19-point margin that made the split verdict look more cautious than the result warranted.

Sacramento was a firmer read. Sacramento Kings vs Los Angeles Lakers prediction saw 70% of the panel side with the Lakers, again at an average confidence of 36%, and the most common scoreline came in at 118-119 — effectively a coin-flip margin on paper. The Lakers won 127-103, a 24-point gap that the models had not anticipated even while picking the right side.

Two correct calls from three is a solid return on a night when the panel never exceeded 36% average confidence on any fixture. Neither winning pick was backed by a large majority, and neither projected scoreline came close to the final margin, which is worth noting before the 2 of 3 record is read as a strong signal.

Where it got it wrong

Charlotte was the day's clear miss, and it was unanimous. Charlotte Hornets vs Brooklyn Nets prediction drew 100% of the ten models behind the Hornets, with the most common scoreline forecast at 112-105 and an average confidence of 36%. Brooklyn won 124-90.

The 34-point margin is the part that stings. A unanimous panel did not merely pick the wrong winner; it picked a side that was beaten by more than 30 points, in a game where the projected four-point edge pointed the other way. The low average confidence of 36% offers some mitigation — the models were not loudly certain — but a 10-of-10 split leaves no room to argue the panel saw the risk coming.

Models of the day

Five models finished Tuesday at 2 of 3: GPT-5.6 Luna, Mistral Medium 3.5, DeepSeek V4 Pro, MiMo V2.5 Pro and Kimi K3. Each of them avoided the Charlotte trap or absorbed it while still landing the other two fixtures.

Three models went 1 of 3: Qwen 3.8 Max, Grok 4.6 and Gemini 3.8 Flash. Two models finished the day without a single correct call: Claude Sonnet 5 and GLM 5.3, both at 0 of 3. On a three-game slate that included one unanimous failure, the spread between the best and worst performers was the full width of the card.

Full results

• NBA: Milwaukee Bucks 97–116 Minnesota Timberwolves — consensus Minnesota Timberwolves win (60%) RIGHT
• NBA: Sacramento Kings 103–127 Los Angeles Lakers — consensus Los Angeles Lakers win (70%) RIGHT
• NBA: Charlotte Hornets 90–124 Brooklyn Nets — consensus Charlotte Hornets win (100%) WRONG

The models return to the board today. Full picks, projected scorelines and confidence levels for the next slate are published on the today's NBA predictions page.

Next up: today's nba predictions from the same panel, and the results archive with every graded call. Forecasts are informational only.