feat: update IteraCompute model catalog

providers/iteracompute/models/deepseek/deepseek-v4-flash-0731.toml

@@ -0,0 +1,16 @@
+# Sources (accessed 2026-09-15):
+# https://api.iteracompute.com/v1/models (model ID, availability, capabilities, limits, and pricing)
+# https://iteracompute.com/docs.html (OpenAI-compatible API documentation)
+# Reasoning is mandatory and the public catalog exposes no caller-selectable reasoning control.
+# Prices are in USD per million tokens.
+base_model = "deepseek/deepseek-v4-flash-0731"
+reasoning_options = []
+
+[cost]
+input = 0.34
+output = 1.05
+cache_read = 0.035
+
+[limit]
+context = 970_000
+output = 393_216

providers/iteracompute/models/deepseek/deepseek-v4-pro-0813.toml

@@ -0,0 +1,16 @@
+# Sources (accessed 2026-09-15):
+# https://api.iteracompute.com/v1/models (model ID, availability, capabilities, limits, and pricing)
+# https://iteracompute.com/docs.html (OpenAI-compatible API documentation)
+# Reasoning is mandatory and the public catalog exposes no caller-selectable reasoning control.
+# Prices are in USD per million tokens.
+base_model = "deepseek/deepseek-v4-pro-0813"
+reasoning_options = []
+
+[cost]
+input = 1.10
+output = 3.30
+cache_read = 0.11
+
+[limit]
+context = 1_048_576
+output = 393_216

providers/iteracompute/models/minimax/minimax-m3.toml

@@ -0,0 +1,19 @@
+# Sources (accessed 2026-09-15):
+# https://api.iteracompute.com/v1/models (model ID, availability, capabilities, limits, and pricing)
+# https://iteracompute.com/docs.html (OpenAI-compatible API documentation)
+# Reasoning is mandatory and the public catalog exposes no caller-selectable reasoning control.
+# Prices are in USD per million tokens.
+base_model = "minimax/MiniMax-M3"
+reasoning_options = []
+
+[cost]
+input = 0.29
+output = 1.20
+cache_read = 0.08
+
+[limit]
+context = 1_048_576
+output = 524_288
+
+[modalities]
+input = ["text", "image"]

providers/iteracompute/models/moonshotai/kimi-k3.toml

@@ -0,0 +1,20 @@
+# Sources (accessed 2026-09-15):
+# https://api.iteracompute.com/v1/models (model ID, availability, capabilities, limits, and pricing)
+# https://iteracompute.com/docs.html (OpenAI-compatible API documentation)
+# Reasoning is mandatory and the public catalog exposes no caller-selectable reasoning control.
+# Prices are in USD per million tokens.
+base_model = "moonshotai/kimi-k3"
+structured_output = false
+reasoning_options = []
+
+[cost]
+input = 3.00
+output = 14.90
+cache_read = 0.29
+
+[limit]
+context = 1_048_576
+output = 999_999
+
+[modalities]
+input = ["text", "image"]

providers/iteracompute/models/ornith-ai/ornith-1.5-35b-a3b.toml

@@ -0,0 +1,22 @@
+# Sources (accessed 2026-09-15):
+# https://api.iteracompute.com/v1/models (canonical model ID, availability, capabilities, limits, and pricing)
+# https://iteracompute.com/docs.html (OpenAI-compatible API documentation)
+# Toggle wire path: reasoning_effort = "none" disables thinking; responses expose reasoning_content.
+# Prices are in USD per million tokens.
+base_model = "deepreinforce/ornith-1.5-35b-a3b"
+
+[interleaved]
+field = "reasoning_content"
+
+[[reasoning_options]]
+type = "toggle"
+
+[cost]
+input = 0.30
+output = 3.00
+cache_read = 0.03
+
+[limit]
+context = 327_680
+input = 262_144
+output = 65_536

providers/iteracompute/models/qwen/qwen3.8-2.4t-a95b.toml

@@ -0,0 +1,19 @@
+# Sources (accessed 2026-09-15):
+# https://api.iteracompute.com/v1/models (model ID, availability, capabilities, limits, and pricing)
+# https://iteracompute.com/docs.html (OpenAI-compatible API documentation)
+# Prices are in USD per million tokens.
+base_model = "alibaba/qwen3.8-2.4t-a95b"
+attachment = true
+reasoning_options = [{ type = "toggle" }]
+
+[cost]
+input = 1.95
+output = 5.95
+cache_read = 0.20
+
+[limit]
+context = 970_000
+output = 131_072
+
+[modalities]
+input = ["text", "image"]

providers/iteracompute/models/qwen/qwen3.8-27b.toml

@@ -0,0 +1,19 @@
+# Sources (accessed 2026-09-15):
+# https://api.iteracompute.com/v1/models (canonical model ID, availability, capabilities, limits, and pricing)
+# https://iteracompute.com/docs.html (OpenAI-compatible API documentation)
+# Prices are in USD per million tokens.
+base_model = "alibaba/qwen3.8-27b"
+reasoning_options = [{ type = "effort", values = ["none", "low", "medium", "xhigh"] }]
+
+[cost]
+input = 0.30
+output = 2.50
+cache_read = 0.03
+
+[limit]
+context = 327_680
+input = 262_144
+output = 65_536
+
+[modalities]
+input = ["text", "image"]

providers/iteracompute/models/z-ai/glm-5.3-flash.toml

@@ -0,0 +1,20 @@
+# Sources (accessed 2026-09-15):
+# https://api.iteracompute.com/v1/models (model ID, availability, capabilities, limits, pricing, and capacity)
+# https://iteracompute.com/docs.html (OpenAI-compatible API documentation)
+# Effort wire path: reasoning.effort = low|high|max. Reasoning is mandatory.
+# Prices are in USD per million tokens.
+base_model = "zhipuai/glm-5.3-flash"
+structured_output = false
+reasoning_options = [{ type = "effort", values = ["low", "high", "max"] }]
+
+[cost]
+input = 0.14
+output = 0.49
+cache_read = 0.03
+
+[limit]
+context = 1_048_576
+output = 131_072
+
+[modalities]
+input = ["text", "image"]

providers/iteracompute/models/z-ai/glm-5.3.toml

@@ -0,0 +1,17 @@
+# Sources (accessed 2026-09-15):
+# https://api.iteracompute.com/v1/models (model ID, availability, capabilities, limits, and pricing)
+# https://iteracompute.com/docs.html (OpenAI-compatible API documentation)
+# Reasoning is mandatory and the public catalog exposes no caller-selectable reasoning control.
+# Prices are in USD per million tokens.
+base_model = "zhipuai/glm-5.3"
+structured_output = false
+reasoning_options = []
+
+[cost]
+input = 1.20
+output = 3.50
+cache_read = 0.26
+
+[limit]
+context = 1_048_576
+output = 131_072