GLM-5.2 Input Price Jumps 675% as Cursor, Qwen Code Flip Rivals

Prices Biggest mover: z-ai/glm-5.2 input up $0.098→$0.76/MTok (+675.5%), output $0.308→$2.42 (evt 50743, 50288). z-ai/glm-5.1 input up $0.952→$1.4/MTok (+47.1%) (evt 50732, 50311). moonshotai/kimi-k2.6 input up $0.5795→$0.95/MTok (+63.9%), output $2.44→$4 (evt 50749, 50309). google/gemma-4-26b-a4b-it input up $0.07→$0.12/MTok (+71.4%) (evt 50685, 50314). deepseek/deepseek-v3.2 input up $0.26→$0.269/MTok (+3.5%) (evt 50726, 50331). On the discount side: nvidia/nemotron-3-super-120b-a12b input cut $0.3→$0.085/MTok (-71.7%) (evt 50680, 50321); qwen/qwen3-14b input cut $0.2275→$0.12/MTok (-47.3%) (evt 50699, 50353); meta-llama/llama-4-maverick output cut $0.8→$0.696/MTok (-13%) (evt 50763, 50354); gryphe/mythomax-l2-13b input cut $0.08→$0.06/MTok (-25%) (evt 50718, 50360); digitalocean/deepseek-4-flash input cut $0.084→$0.0679/MTok (-19.2%) (evt 50551).

Usage Cursor overtook Codebuff on OpenRouter coding usage, 129,513,972,577 vs 117,125,611,935 tokens (7d, +10.6%) (evt 51115). Qwen Code overtook Letaido, 23,348,175,309 vs 22,122,601,914 tokens (7d, +5.5%) (evt 51116). gpt-oss-120b passed 50B OpenRouter tokens/day at 57,717,627,882 (evt 52240).

Releases qwen-code shipped v0.21.8-nightly.20260810.55e20db328 (evt 51298); opencode v1.18.16 (evt 51297); gemini-cli v0.56.0-nightly.20260810.gcf22ac7e8 (evt 51296).

Plans Codex removed Pro 5x/20x tiers, added a Pro plan, and dropped its listed Free price (evt 51107). Base44 relisted Elite/Pro/Builder/Starter as monthly prices ($160/$80/$40/$16) and removed Enterprise (evt 51106). Lovable listed Pro at $25/mo and Business at $50/mo (evt 51112).